block/cfq-iosched.c

   1 /*
   2  *  CFQ, or complete fairness queueing, disk scheduler.
   3  *
   4  *  Based on ideas from a previously unfinished io
   5  *  scheduler (round robin per-process disk scheduling) and Andrea Arcangeli.
   6  *
   7  *  Copyright (C) 2003 Jens Axboe <axboe@kernel.dk>
   8  */
   9 #include <linux/module.h>
  10 #include <linux/slab.h>
  11 #include <linux/blkdev.h>
  12 #include <linux/elevator.h>
  13 #include <linux/ktime.h>
  14 #include <linux/rbtree.h>
  15 #include <linux/ioprio.h>
  16 #include <linux/blktrace_api.h>
  17 #include <linux/blk-cgroup.h>
  18 #include "blk.h"
  19 #include "blk-wbt.h"
  20
  21 /*
  22  * tunables
  23  */
  24 /* max queue in one round of service */
  25 static const int cfq_quantum = 8;
  26 static const u64 cfq_fifo_expire[2] = { NSEC_PER_SEC / 4, NSEC_PER_SEC / 8 };
  27 /* maximum backwards seek, in KiB */
  28 static const int cfq_back_max = 16 * 1024;
  29 /* penalty of a backwards seek */
  30 static const int cfq_back_penalty = 2;
  31 static const u64 cfq_slice_sync = NSEC_PER_SEC / 10;
  32 static u64 cfq_slice_async = NSEC_PER_SEC / 25;
  33 static const int cfq_slice_async_rq = 2;
  34 static u64 cfq_slice_idle = NSEC_PER_SEC / 125;
  35 static u64 cfq_group_idle = NSEC_PER_SEC / 125;
  36 static const u64 cfq_target_latency = (u64)NSEC_PER_SEC * 3/10; /* 300 ms */
  37 static const int cfq_hist_divisor = 4;
  38
  39 /*
  40  * offset from end of service tree
  41  */
  42 #define CFQ_IDLE_DELAY          (NSEC_PER_SEC / 5)
  43
  44 /*
  45  * below this threshold, we consider thinktime immediate
  46  */
  47 #define CFQ_MIN_TT              (2 * NSEC_PER_SEC / HZ)
  48
  49 #define CFQ_SLICE_SCALE         (5)
  50 #define CFQ_HW_QUEUE_MIN        (5)
  51 #define CFQ_SERVICE_SHIFT       12
  52
  53 #define CFQQ_SEEK_THR           (sector_t)(8 * 100)
  54 #define CFQQ_CLOSE_THR          (sector_t)(8 * 1024)
  55 #define CFQQ_SECT_THR_NONROT    (sector_t)(2 * 32)
  56 #define CFQQ_SEEKY(cfqq)        (hweight32(cfqq->seek_history) > 32/8)
  57
  58 #define RQ_CIC(rq)              icq_to_cic((rq)->elv.icq)
  59 #define RQ_CFQQ(rq)             (struct cfq_queue *) ((rq)->elv.priv[0])
  60 #define RQ_CFQG(rq)             (struct cfq_group *) ((rq)->elv.priv[1])
  61
  62 static struct kmem_cache *cfq_pool;
  63
  64 #define CFQ_PRIO_LISTS          IOPRIO_BE_NR
  65 #define cfq_class_idle(cfqq)    ((cfqq)->ioprio_class == IOPRIO_CLASS_IDLE)
  66 #define cfq_class_rt(cfqq)      ((cfqq)->ioprio_class == IOPRIO_CLASS_RT)
  67
  68 #define sample_valid(samples)   ((samples) > 80)
  69 #define rb_entry_cfqg(node)     rb_entry((node), struct cfq_group, rb_node)
  70
  71 /* blkio-related constants */
  72 #define CFQ_WEIGHT_LEGACY_MIN   10
  73 #define CFQ_WEIGHT_LEGACY_DFL   500
  74 #define CFQ_WEIGHT_LEGACY_MAX   1000
  75
  76 struct cfq_ttime {
  77         u64 last_end_request;
  78
  79         u64 ttime_total;
  80         u64 ttime_mean;
  81         unsigned long ttime_samples;
  82 };
  83
  84 /*
  85  * Most of our rbtree usage is for sorting with min extraction, so
  86  * if we cache the leftmost node we don't have to walk down the tree
  87  * to find it. Idea borrowed from Ingo Molnars CFS scheduler. We should
  88  * move this into the elevator for the rq sorting as well.
  89  */
  90 struct cfq_rb_root {
  91         struct rb_root rb;
  92         struct rb_node *left;
  93         unsigned count;
  94         u64 min_vdisktime;
  95         struct cfq_ttime ttime;
  96 };
  97 #define CFQ_RB_ROOT     (struct cfq_rb_root) { .rb = RB_ROOT, \
  98                         .ttime = {.last_end_request = ktime_get_ns(),},}
  99
 100 /*
 101  * Per process-grouping structure
 102  */
 103 struct cfq_queue {
 104         /* reference count */
 105         int ref;
 106         /* various state flags, see below */
 107         unsigned int flags;
 108         /* parent cfq_data */
 109         struct cfq_data *cfqd;
 110         /* service_tree member */
 111         struct rb_node rb_node;
 112         /* service_tree key */
 113         u64 rb_key;
 114         /* prio tree member */
 115         struct rb_node p_node;
 116         /* prio tree root we belong to, if any */
 117         struct rb_root *p_root;
 118         /* sorted list of pending requests */
 119         struct rb_root sort_list;
 120         /* if fifo isn't expired, next request to serve */
 121         struct request *next_rq;
 122         /* requests queued in sort_list */
 123         int queued[2];
 124         /* currently allocated requests */
 125         int allocated[2];
 126         /* fifo list of requests in sort_list */
 127         struct list_head fifo;
 128
 129         /* time when queue got scheduled in to dispatch first request. */
 130         u64 dispatch_start;
 131         u64 allocated_slice;
 132         u64 slice_dispatch;
 133         /* time when first request from queue completed and slice started. */
 134         u64 slice_start;
 135         u64 slice_end;
 136         s64 slice_resid;
 137
 138         /* pending priority requests */
 139         int prio_pending;
 140         /* number of requests that are on the dispatch list or inside driver */
 141         int dispatched;
 142
 143         /* io prio of this group */
 144         unsigned short ioprio, org_ioprio;
 145         unsigned short ioprio_class, org_ioprio_class;
 146
 147         pid_t pid;
 148
 149         u32 seek_history;
 150         sector_t last_request_pos;
 151
 152         struct cfq_rb_root *service_tree;
 153         struct cfq_queue *new_cfqq;
 154         struct cfq_group *cfqg;
 155         /* Number of sectors dispatched from queue in single dispatch round */
 156         unsigned long nr_sectors;
 157 };
 158
 159 /*
 160  * First index in the service_trees.
 161  * IDLE is handled separately, so it has negative index
 162  */
 163 enum wl_class_t {
 164         BE_WORKLOAD = 0,
 165         RT_WORKLOAD = 1,
 166         IDLE_WORKLOAD = 2,
 167         CFQ_PRIO_NR,
 168 };
 169
 170 /*
 171  * Second index in the service_trees.
 172  */
 173 enum wl_type_t {
 174         ASYNC_WORKLOAD = 0,
 175         SYNC_NOIDLE_WORKLOAD = 1,
 176         SYNC_WORKLOAD = 2
 177 };
 178
 179 struct cfqg_stats {
 180 #ifdef CONFIG_CFQ_GROUP_IOSCHED
 181         /* number of ios merged */
 182         struct blkg_rwstat              merged;
 183         /* total time spent on device in ns, may not be accurate w/ queueing */
 184         struct blkg_rwstat              service_time;
 185         /* total time spent waiting in scheduler queue in ns */
 186         struct blkg_rwstat              wait_time;
 187         /* number of IOs queued up */
 188         struct blkg_rwstat              queued;
 189         /* total disk time and nr sectors dispatched by this group */
 190         struct blkg_stat                time;
 191 #ifdef CONFIG_DEBUG_BLK_CGROUP
 192         /* time not charged to this cgroup */
 193         struct blkg_stat                unaccounted_time;
 194         /* sum of number of ios queued across all samples */
 195         struct blkg_stat                avg_queue_size_sum;
 196         /* count of samples taken for average */
 197         struct blkg_stat                avg_queue_size_samples;
 198         /* how many times this group has been removed from service tree */
 199         struct blkg_stat                dequeue;
 200         /* total time spent waiting for it to be assigned a timeslice. */
 201         struct blkg_stat                group_wait_time;
 202         /* time spent idling for this blkcg_gq */
 203         struct blkg_stat                idle_time;
 204         /* total time with empty current active q with other requests queued */
 205         struct blkg_stat                empty_time;
 206         /* fields after this shouldn't be cleared on stat reset */
 207         uint64_t                        start_group_wait_time;
 208         uint64_t                        start_idle_time;
 209         uint64_t                        start_empty_time;
 210         uint16_t                        flags;
 211 #endif  /* CONFIG_DEBUG_BLK_CGROUP */
 212 #endif  /* CONFIG_CFQ_GROUP_IOSCHED */
 213 };
 214
 215 /* Per-cgroup data */
 216 struct cfq_group_data {
 217         /* must be the first member */
 218         struct blkcg_policy_data cpd;
 219
 220         unsigned int weight;
 221         unsigned int leaf_weight;
 222 };
 223
 224 /* This is per cgroup per device grouping structure */
 225 struct cfq_group {
 226         /* must be the first member */
 227         struct blkg_policy_data pd;
 228
 229         /* group service_tree member */
 230         struct rb_node rb_node;
 231
 232         /* group service_tree key */
 233         u64 vdisktime;
 234
 235         /*
 236          * The number of active cfqgs and sum of their weights under this
 237          * cfqg.  This covers this cfqg's leaf_weight and all children's
 238          * weights, but does not cover weights of further descendants.
 239          *
 240          * If a cfqg is on the service tree, it's active.  An active cfqg
 241          * also activates its parent and contributes to the children_weight
 242          * of the parent.
 243          */
 244         int nr_active;
 245         unsigned int children_weight;
 246
 247         /*
 248          * vfraction is the fraction of vdisktime that the tasks in this
 249          * cfqg are entitled to.  This is determined by compounding the
 250          * ratios walking up from this cfqg to the root.
 251          *
 252          * It is in fixed point w/ CFQ_SERVICE_SHIFT and the sum of all
 253          * vfractions on a service tree is approximately 1.  The sum may
 254          * deviate a bit due to rounding errors and fluctuations caused by
 255          * cfqgs entering and leaving the service tree.
 256          */
 257         unsigned int vfraction;
 258
 259         /*
 260          * There are two weights - (internal) weight is the weight of this
 261          * cfqg against the sibling cfqgs.  leaf_weight is the wight of
 262          * this cfqg against the child cfqgs.  For the root cfqg, both
 263          * weights are kept in sync for backward compatibility.
 264          */
 265         unsigned int weight;
 266         unsigned int new_weight;
 267         unsigned int dev_weight;
 268
 269         unsigned int leaf_weight;
 270         unsigned int new_leaf_weight;
 271         unsigned int dev_leaf_weight;
 272
 273         /* number of cfqq currently on this group */
 274         int nr_cfqq;
 275
 276         /*
 277          * Per group busy queues average. Useful for workload slice calc. We
 278          * create the array for each prio class but at run time it is used
 279          * only for RT and BE class and slot for IDLE class remains unused.
 280          * This is primarily done to avoid confusion and a gcc warning.
 281          */
 282         unsigned int busy_queues_avg[CFQ_PRIO_NR];
 283         /*
 284          * rr lists of queues with requests. We maintain service trees for
 285          * RT and BE classes. These trees are subdivided in subclasses
 286          * of SYNC, SYNC_NOIDLE and ASYNC based on workload type. For IDLE
 287          * class there is no subclassification and all the cfq queues go on
 288          * a single tree service_tree_idle.
 289          * Counts are embedded in the cfq_rb_root
 290          */
 291         struct cfq_rb_root service_trees[2][3];
 292         struct cfq_rb_root service_tree_idle;
 293
 294         u64 saved_wl_slice;
 295         enum wl_type_t saved_wl_type;
 296         enum wl_class_t saved_wl_class;
 297
 298         /* number of requests that are on the dispatch list or inside driver */
 299         int dispatched;
 300         struct cfq_ttime ttime;
 301         struct cfqg_stats stats;        /* stats for this cfqg */
 302
 303         /* async queue for each priority case */
 304         struct cfq_queue *async_cfqq[2][IOPRIO_BE_NR];
 305         struct cfq_queue *async_idle_cfqq;
 306
 307 };
 308
 309 struct cfq_io_cq {
 310         struct io_cq            icq;            /* must be the first member */
 311         struct cfq_queue        *cfqq[2];
 312         struct cfq_ttime        ttime;
 313         int                     ioprio;         /* the current ioprio */
 314 #ifdef CONFIG_CFQ_GROUP_IOSCHED
 315         uint64_t                blkcg_serial_nr; /* the current blkcg serial */
 316 #endif
 317 };
 318
 319 /*
 320  * Per block device queue structure
 321  */
 322 struct cfq_data {
 323         struct request_queue *queue;
 324         /* Root service tree for cfq_groups */
 325         struct cfq_rb_root grp_service_tree;
 326         struct cfq_group *root_group;
 327
 328         /*
 329          * The priority currently being served
 330          */
 331         enum wl_class_t serving_wl_class;
 332         enum wl_type_t serving_wl_type;
 333         u64 workload_expires;
 334         struct cfq_group *serving_group;
 335
 336         /*
 337          * Each priority tree is sorted by next_request position.  These
 338          * trees are used when determining if two or more queues are
 339          * interleaving requests (see cfq_close_cooperator).
 340          */
 341         struct rb_root prio_trees[CFQ_PRIO_LISTS];
 342
 343         unsigned int busy_queues;
 344         unsigned int busy_sync_queues;
 345
 346         int rq_in_driver;
 347         int rq_in_flight[2];
 348
 349         /*
 350          * queue-depth detection
 351          */
 352         int rq_queued;
 353         int hw_tag;
 354         /*
 355          * hw_tag can be
 356          * -1 => indeterminate, (cfq will behave as if NCQ is present, to allow better detection)
 357          *  1 => NCQ is present (hw_tag_est_depth is the estimated max depth)
 358          *  0 => no NCQ
 359          */
 360         int hw_tag_est_depth;
 361         unsigned int hw_tag_samples;
 362
 363         /*
 364          * idle window management
 365          */
 366         struct hrtimer idle_slice_timer;
 367         struct work_struct unplug_work;
 368
 369         struct cfq_queue *active_queue;
 370         struct cfq_io_cq *active_cic;
 371
 372         sector_t last_position;
 373
 374         /*
 375          * tunables, see top of file
 376          */
 377         unsigned int cfq_quantum;
 378         unsigned int cfq_back_penalty;
 379         unsigned int cfq_back_max;
 380         unsigned int cfq_slice_async_rq;
 381         unsigned int cfq_latency;
 382         u64 cfq_fifo_expire[2];
 383         u64 cfq_slice[2];
 384         u64 cfq_slice_idle;
 385         u64 cfq_group_idle;
 386         u64 cfq_target_latency;
 387
 388         /*
 389          * Fallback dummy cfqq for extreme OOM conditions
 390          */
 391         struct cfq_queue oom_cfqq;
 392
 393         u64 last_delayed_sync;
 394 };
 395
 396 static struct cfq_group *cfq_get_next_cfqg(struct cfq_data *cfqd);
 397 static void cfq_put_queue(struct cfq_queue *cfqq);
 398
 399 static struct cfq_rb_root *st_for(struct cfq_group *cfqg,
 400                                             enum wl_class_t class,
 401                                             enum wl_type_t type)
 402 {
 403         if (!cfqg)
 404                 return NULL;
 405
 406         if (class == IDLE_WORKLOAD)
 407                 return &cfqg->service_tree_idle;
 408
 409         return &cfqg->service_trees[class][type];
 410 }
 411
 412 enum cfqq_state_flags {
 413         CFQ_CFQQ_FLAG_on_rr = 0,        /* on round-robin busy list */
 414         CFQ_CFQQ_FLAG_wait_request,     /* waiting for a request */
 415         CFQ_CFQQ_FLAG_must_dispatch,    /* must be allowed a dispatch */
 416         CFQ_CFQQ_FLAG_must_alloc_slice, /* per-slice must_alloc flag */
 417         CFQ_CFQQ_FLAG_fifo_expire,      /* FIFO checked in this slice */
 418         CFQ_CFQQ_FLAG_idle_window,      /* slice idling enabled */
 419         CFQ_CFQQ_FLAG_prio_changed,     /* task priority has changed */
 420         CFQ_CFQQ_FLAG_slice_new,        /* no requests dispatched in slice */
 421         CFQ_CFQQ_FLAG_sync,             /* synchronous queue */
 422         CFQ_CFQQ_FLAG_coop,             /* cfqq is shared */
 423         CFQ_CFQQ_FLAG_split_coop,       /* shared cfqq will be splitted */
 424         CFQ_CFQQ_FLAG_deep,             /* sync cfqq experienced large depth */
 425         CFQ_CFQQ_FLAG_wait_busy,        /* Waiting for next request */
 426 };
 427
 428 #define CFQ_CFQQ_FNS(name)                                              \
 429 static inline void cfq_mark_cfqq_##name(struct cfq_queue *cfqq)         \
 430 {                                                                       \
 431         (cfqq)->flags |= (1 << CFQ_CFQQ_FLAG_##name);                   \
 432 }                                                                       \
 433 static inline void cfq_clear_cfqq_##name(struct cfq_queue *cfqq)        \
 434 {                                                                       \
 435         (cfqq)->flags &= ~(1 << CFQ_CFQQ_FLAG_##name);                  \
 436 }                                                                       \
 437 static inline int cfq_cfqq_##name(const struct cfq_queue *cfqq)         \
 438 {                                                                       \
 439         return ((cfqq)->flags & (1 << CFQ_CFQQ_FLAG_##name)) != 0;      \
 440 }
 441
 442 CFQ_CFQQ_FNS(on_rr);
 443 CFQ_CFQQ_FNS(wait_request);
 444 CFQ_CFQQ_FNS(must_dispatch);
 445 CFQ_CFQQ_FNS(must_alloc_slice);
 446 CFQ_CFQQ_FNS(fifo_expire);
 447 CFQ_CFQQ_FNS(idle_window);
 448 CFQ_CFQQ_FNS(prio_changed);
 449 CFQ_CFQQ_FNS(slice_new);
 450 CFQ_CFQQ_FNS(sync);
 451 CFQ_CFQQ_FNS(coop);
 452 CFQ_CFQQ_FNS(split_coop);
 453 CFQ_CFQQ_FNS(deep);
 454 CFQ_CFQQ_FNS(wait_busy);
 455 #undef CFQ_CFQQ_FNS
 456
 457 #if defined(CONFIG_CFQ_GROUP_IOSCHED) && defined(CONFIG_DEBUG_BLK_CGROUP)
 458
 459 /* cfqg stats flags */
 460 enum cfqg_stats_flags {
 461         CFQG_stats_waiting = 0,
 462         CFQG_stats_idling,
 463         CFQG_stats_empty,
 464 };
 465
 466 #define CFQG_FLAG_FNS(name)                                             \
 467 static inline void cfqg_stats_mark_##name(struct cfqg_stats *stats)     \
 468 {                                                                       \
 469         stats->flags |= (1 << CFQG_stats_##name);                       \
 470 }                                                                       \
 471 static inline void cfqg_stats_clear_##name(struct cfqg_stats *stats)    \
 472 {                                                                       \
 473         stats->flags &= ~(1 << CFQG_stats_##name);                      \
 474 }                                                                       \
 475 static inline int cfqg_stats_##name(struct cfqg_stats *stats)           \
 476 {                                                                       \
 477         return (stats->flags & (1 << CFQG_stats_##name)) != 0;          \
 478 }                                                                       \
 479
 480 CFQG_FLAG_FNS(waiting)
 481 CFQG_FLAG_FNS(idling)
 482 CFQG_FLAG_FNS(empty)
 483 #undef CFQG_FLAG_FNS
 484
 485 /* This should be called with the queue_lock held. */
 486 static void cfqg_stats_update_group_wait_time(struct cfqg_stats *stats)
 487 {
 488         unsigned long long now;
 489
 490         if (!cfqg_stats_waiting(stats))
 491                 return;
 492
 493         now = sched_clock();
 494         if (time_after64(now, stats->start_group_wait_time))
 495                 blkg_stat_add(&stats->group_wait_time,
 496                               now - stats->start_group_wait_time);
 497         cfqg_stats_clear_waiting(stats);
 498 }
 499
 500 /* This should be called with the queue_lock held. */
 501 static void cfqg_stats_set_start_group_wait_time(struct cfq_group *cfqg,
 502                                                  struct cfq_group *curr_cfqg)
 503 {
 504         struct cfqg_stats *stats = &cfqg->stats;
 505
 506         if (cfqg_stats_waiting(stats))
 507                 return;
 508         if (cfqg == curr_cfqg)
 509                 return;
 510         stats->start_group_wait_time = sched_clock();
 511         cfqg_stats_mark_waiting(stats);
 512 }
 513
 514 /* This should be called with the queue_lock held. */
 515 static void cfqg_stats_end_empty_time(struct cfqg_stats *stats)
 516 {
 517         unsigned long long now;
 518
 519         if (!cfqg_stats_empty(stats))
 520                 return;
 521
 522         now = sched_clock();
 523         if (time_after64(now, stats->start_empty_time))
 524                 blkg_stat_add(&stats->empty_time,
 525                               now - stats->start_empty_time);
 526         cfqg_stats_clear_empty(stats);
 527 }
 528
 529 static void cfqg_stats_update_dequeue(struct cfq_group *cfqg)
 530 {
 531         blkg_stat_add(&cfqg->stats.dequeue, 1);
 532 }
 533
 534 static void cfqg_stats_set_start_empty_time(struct cfq_group *cfqg)
 535 {
 536         struct cfqg_stats *stats = &cfqg->stats;
 537
 538         if (blkg_rwstat_total(&stats->queued))
 539                 return;
 540
 541         /*
 542          * group is already marked empty. This can happen if cfqq got new
 543          * request in parent group and moved to this group while being added
 544          * to service tree. Just ignore the event and move on.
 545          */
 546         if (cfqg_stats_empty(stats))
 547                 return;
 548
 549         stats->start_empty_time = sched_clock();
 550         cfqg_stats_mark_empty(stats);
 551 }
 552
 553 static void cfqg_stats_update_idle_time(struct cfq_group *cfqg)
 554 {
 555         struct cfqg_stats *stats = &cfqg->stats;
 556
 557         if (cfqg_stats_idling(stats)) {
 558                 unsigned long long now = sched_clock();
 559
 560                 if (time_after64(now, stats->start_idle_time))
 561                         blkg_stat_add(&stats->idle_time,
 562                                       now - stats->start_idle_time);
 563                 cfqg_stats_clear_idling(stats);
 564         }
 565 }
 566
 567 static void cfqg_stats_set_start_idle_time(struct cfq_group *cfqg)
 568 {
 569         struct cfqg_stats *stats = &cfqg->stats;
 570
 571         BUG_ON(cfqg_stats_idling(stats));
 572
 573         stats->start_idle_time = sched_clock();
 574         cfqg_stats_mark_idling(stats);
 575 }
 576
 577 static void cfqg_stats_update_avg_queue_size(struct cfq_group *cfqg)
 578 {
 579         struct cfqg_stats *stats = &cfqg->stats;
 580
 581         blkg_stat_add(&stats->avg_queue_size_sum,
 582                       blkg_rwstat_total(&stats->queued));
 583         blkg_stat_add(&stats->avg_queue_size_samples, 1);
 584         cfqg_stats_update_group_wait_time(stats);
 585 }
 586
 587 #else   /* CONFIG_CFQ_GROUP_IOSCHED && CONFIG_DEBUG_BLK_CGROUP */
 588
 589 static inline void cfqg_stats_set_start_group_wait_time(struct cfq_group *cfqg, struct cfq_group *curr_cfqg) { }
 590 static inline void cfqg_stats_end_empty_time(struct cfqg_stats *stats) { }
 591 static inline void cfqg_stats_update_dequeue(struct cfq_group *cfqg) { }
 592 static inline void cfqg_stats_set_start_empty_time(struct cfq_group *cfqg) { }
 593 static inline void cfqg_stats_update_idle_time(struct cfq_group *cfqg) { }
 594 static inline void cfqg_stats_set_start_idle_time(struct cfq_group *cfqg) { }
 595 static inline void cfqg_stats_update_avg_queue_size(struct cfq_group *cfqg) { }
 596
 597 #endif  /* CONFIG_CFQ_GROUP_IOSCHED && CONFIG_DEBUG_BLK_CGROUP */
 598
 599 #ifdef CONFIG_CFQ_GROUP_IOSCHED
 600
 601 static inline struct cfq_group *pd_to_cfqg(struct blkg_policy_data *pd)
 602 {
 603         return pd ? container_of(pd, struct cfq_group, pd) : NULL;
 604 }
 605
 606 static struct cfq_group_data
 607 *cpd_to_cfqgd(struct blkcg_policy_data *cpd)
 608 {
 609         return cpd ? container_of(cpd, struct cfq_group_data, cpd) : NULL;
 610 }
 611
 612 static inline struct blkcg_gq *cfqg_to_blkg(struct cfq_group *cfqg)
 613 {
 614         return pd_to_blkg(&cfqg->pd);
 615 }
 616
 617 static struct blkcg_policy blkcg_policy_cfq;
 618
 619 static inline struct cfq_group *blkg_to_cfqg(struct blkcg_gq *blkg)
 620 {
 621         return pd_to_cfqg(blkg_to_pd(blkg, &blkcg_policy_cfq));
 622 }
 623
 624 static struct cfq_group_data *blkcg_to_cfqgd(struct blkcg *blkcg)
 625 {
 626         return cpd_to_cfqgd(blkcg_to_cpd(blkcg, &blkcg_policy_cfq));
 627 }
 628
 629 static inline struct cfq_group *cfqg_parent(struct cfq_group *cfqg)
 630 {
 631         struct blkcg_gq *pblkg = cfqg_to_blkg(cfqg)->parent;
 632
 633         return pblkg ? blkg_to_cfqg(pblkg) : NULL;
 634 }
 635
 636 static inline bool cfqg_is_descendant(struct cfq_group *cfqg,
 637                                       struct cfq_group *ancestor)
 638 {
 639         return cgroup_is_descendant(cfqg_to_blkg(cfqg)->blkcg->css.cgroup,
 640                                     cfqg_to_blkg(ancestor)->blkcg->css.cgroup);
 641 }
 642
 643 static inline void cfqg_get(struct cfq_group *cfqg)
 644 {
 645         return blkg_get(cfqg_to_blkg(cfqg));
 646 }
 647
 648 static inline void cfqg_put(struct cfq_group *cfqg)
 649 {
 650         return blkg_put(cfqg_to_blkg(cfqg));
 651 }
 652
 653 #define cfq_log_cfqq(cfqd, cfqq, fmt, args...)  do {                    \
 654         char __pbuf[128];                                               \
 655                                                                         \
 656         blkg_path(cfqg_to_blkg((cfqq)->cfqg), __pbuf, sizeof(__pbuf));  \
 657         blk_add_trace_msg((cfqd)->queue, "cfq%d%c%c %s " fmt, (cfqq)->pid, \
 658                         cfq_cfqq_sync((cfqq)) ? 'S' : 'A',              \
 659                         cfqq_type((cfqq)) == SYNC_NOIDLE_WORKLOAD ? 'N' : ' ',\
 660                           __pbuf, ##args);                              \
 661 } while (0)
 662
 663 #define cfq_log_cfqg(cfqd, cfqg, fmt, args...)  do {                    \
 664         char __pbuf[128];                                               \
 665                                                                         \
 666         blkg_path(cfqg_to_blkg(cfqg), __pbuf, sizeof(__pbuf));          \
 667         blk_add_trace_msg((cfqd)->queue, "%s " fmt, __pbuf, ##args);    \
 668 } while (0)
 669
 670 static inline void cfqg_stats_update_io_add(struct cfq_group *cfqg,
 671                                             struct cfq_group *curr_cfqg,
 672                                             unsigned int op)
 673 {
 674         blkg_rwstat_add(&cfqg->stats.queued, op, 1);
 675         cfqg_stats_end_empty_time(&cfqg->stats);
 676         cfqg_stats_set_start_group_wait_time(cfqg, curr_cfqg);
 677 }
 678
 679 static inline void cfqg_stats_update_timeslice_used(struct cfq_group *cfqg,
 680                         uint64_t time, unsigned long unaccounted_time)
 681 {
 682         blkg_stat_add(&cfqg->stats.time, time);
 683 #ifdef CONFIG_DEBUG_BLK_CGROUP
 684         blkg_stat_add(&cfqg->stats.unaccounted_time, unaccounted_time);
 685 #endif
 686 }
 687
 688 static inline void cfqg_stats_update_io_remove(struct cfq_group *cfqg,
 689                                                unsigned int op)
 690 {
 691         blkg_rwstat_add(&cfqg->stats.queued, op, -1);
 692 }
 693
 694 static inline void cfqg_stats_update_io_merged(struct cfq_group *cfqg,
 695                                                unsigned int op)
 696 {
 697         blkg_rwstat_add(&cfqg->stats.merged, op, 1);
 698 }
 699
 700 static inline void cfqg_stats_update_completion(struct cfq_group *cfqg,
 701                         uint64_t start_time, uint64_t io_start_time,
 702                         unsigned int op)
 703 {
 704         struct cfqg_stats *stats = &cfqg->stats;
 705         unsigned long long now = sched_clock();
 706
 707         if (time_after64(now, io_start_time))
 708                 blkg_rwstat_add(&stats->service_time, op, now - io_start_time);
 709         if (time_after64(io_start_time, start_time))
 710                 blkg_rwstat_add(&stats->wait_time, op,
 711                                 io_start_time - start_time);
 712 }
 713
 714 /* @stats = 0 */
 715 static void cfqg_stats_reset(struct cfqg_stats *stats)
 716 {
 717         /* queued stats shouldn't be cleared */
 718         blkg_rwstat_reset(&stats->merged);
 719         blkg_rwstat_reset(&stats->service_time);
 720         blkg_rwstat_reset(&stats->wait_time);
 721         blkg_stat_reset(&stats->time);
 722 #ifdef CONFIG_DEBUG_BLK_CGROUP
 723         blkg_stat_reset(&stats->unaccounted_time);
 724         blkg_stat_reset(&stats->avg_queue_size_sum);
 725         blkg_stat_reset(&stats->avg_queue_size_samples);
 726         blkg_stat_reset(&stats->dequeue);
 727         blkg_stat_reset(&stats->group_wait_time);
 728         blkg_stat_reset(&stats->idle_time);
 729         blkg_stat_reset(&stats->empty_time);
 730 #endif
 731 }
 732
 733 /* @to += @from */
 734 static void cfqg_stats_add_aux(struct cfqg_stats *to, struct cfqg_stats *from)
 735 {
 736         /* queued stats shouldn't be cleared */
 737         blkg_rwstat_add_aux(&to->merged, &from->merged);
 738         blkg_rwstat_add_aux(&to->service_time, &from->service_time);
 739         blkg_rwstat_add_aux(&to->wait_time, &from->wait_time);
 740         blkg_stat_add_aux(&from->time, &from->time);
 741 #ifdef CONFIG_DEBUG_BLK_CGROUP
 742         blkg_stat_add_aux(&to->unaccounted_time, &from->unaccounted_time);
 743         blkg_stat_add_aux(&to->avg_queue_size_sum, &from->avg_queue_size_sum);
 744         blkg_stat_add_aux(&to->avg_queue_size_samples, &from->avg_queue_size_samples);
 745         blkg_stat_add_aux(&to->dequeue, &from->dequeue);
 746         blkg_stat_add_aux(&to->group_wait_time, &from->group_wait_time);
 747         blkg_stat_add_aux(&to->idle_time, &from->idle_time);
 748         blkg_stat_add_aux(&to->empty_time, &from->empty_time);
 749 #endif
 750 }
 751
 752 /*
 753  * Transfer @cfqg's stats to its parent's aux counts so that the ancestors'
 754  * recursive stats can still account for the amount used by this cfqg after
 755  * it's gone.
 756  */
 757 static void cfqg_stats_xfer_dead(struct cfq_group *cfqg)
 758 {
 759         struct cfq_group *parent = cfqg_parent(cfqg);
 760
 761         lockdep_assert_held(cfqg_to_blkg(cfqg)->q->queue_lock);
 762
 763         if (unlikely(!parent))
 764                 return;
 765
 766         cfqg_stats_add_aux(&parent->stats, &cfqg->stats);
 767         cfqg_stats_reset(&cfqg->stats);
 768 }
 769
 770 #else   /* CONFIG_CFQ_GROUP_IOSCHED */
 771
 772 static inline struct cfq_group *cfqg_parent(struct cfq_group *cfqg) { return NULL; }
 773 static inline bool cfqg_is_descendant(struct cfq_group *cfqg,
 774                                       struct cfq_group *ancestor)
 775 {
 776         return true;
 777 }
 778 static inline void cfqg_get(struct cfq_group *cfqg) { }
 779 static inline void cfqg_put(struct cfq_group *cfqg) { }
 780
 781 #define cfq_log_cfqq(cfqd, cfqq, fmt, args...)  \
 782         blk_add_trace_msg((cfqd)->queue, "cfq%d%c%c " fmt, (cfqq)->pid, \
 783                         cfq_cfqq_sync((cfqq)) ? 'S' : 'A',              \
 784                         cfqq_type((cfqq)) == SYNC_NOIDLE_WORKLOAD ? 'N' : ' ',\
 785                                 ##args)
 786 #define cfq_log_cfqg(cfqd, cfqg, fmt, args...)          do {} while (0)
 787
 788 static inline void cfqg_stats_update_io_add(struct cfq_group *cfqg,
 789                         struct cfq_group *curr_cfqg, unsigned int op) { }
 790 static inline void cfqg_stats_update_timeslice_used(struct cfq_group *cfqg,
 791                         uint64_t time, unsigned long unaccounted_time) { }
 792 static inline void cfqg_stats_update_io_remove(struct cfq_group *cfqg,
 793                         unsigned int op) { }
 794 static inline void cfqg_stats_update_io_merged(struct cfq_group *cfqg,
 795                         unsigned int op) { }
 796 static inline void cfqg_stats_update_completion(struct cfq_group *cfqg,
 797                         uint64_t start_time, uint64_t io_start_time,
 798                         unsigned int op) { }
 799
 800 #endif  /* CONFIG_CFQ_GROUP_IOSCHED */
 801
 802 #define cfq_log(cfqd, fmt, args...)     \
 803         blk_add_trace_msg((cfqd)->queue, "cfq " fmt, ##args)
 804
 805 /* Traverses through cfq group service trees */
 806 #define for_each_cfqg_st(cfqg, i, j, st) \
 807         for (i = 0; i <= IDLE_WORKLOAD; i++) \
 808                 for (j = 0, st = i < IDLE_WORKLOAD ? &cfqg->service_trees[i][j]\
 809                         : &cfqg->service_tree_idle; \
 810                         (i < IDLE_WORKLOAD && j <= SYNC_WORKLOAD) || \
 811                         (i == IDLE_WORKLOAD && j == 0); \
 812                         j++, st = i < IDLE_WORKLOAD ? \
 813                         &cfqg->service_trees[i][j]: NULL) \
 814
 815 static inline bool cfq_io_thinktime_big(struct cfq_data *cfqd,
 816         struct cfq_ttime *ttime, bool group_idle)
 817 {
 818         u64 slice;
 819         if (!sample_valid(ttime->ttime_samples))
 820                 return false;
 821         if (group_idle)
 822                 slice = cfqd->cfq_group_idle;
 823         else
 824                 slice = cfqd->cfq_slice_idle;
 825         return ttime->ttime_mean > slice;
 826 }
 827
 828 static inline bool iops_mode(struct cfq_data *cfqd)
 829 {
 830         /*
 831          * If we are not idling on queues and it is a NCQ drive, parallel
 832          * execution of requests is on and measuring time is not possible
 833          * in most of the cases until and unless we drive shallower queue
 834          * depths and that becomes a performance bottleneck. In such cases
 835          * switch to start providing fairness in terms of number of IOs.
 836          */
 837         if (!cfqd->cfq_slice_idle && cfqd->hw_tag)
 838                 return true;
 839         else
 840                 return false;
 841 }
 842
 843 static inline enum wl_class_t cfqq_class(struct cfq_queue *cfqq)
 844 {
 845         if (cfq_class_idle(cfqq))
 846                 return IDLE_WORKLOAD;
 847         if (cfq_class_rt(cfqq))
 848                 return RT_WORKLOAD;
 849         return BE_WORKLOAD;
 850 }
 851
 852
 853 static enum wl_type_t cfqq_type(struct cfq_queue *cfqq)
 854 {
 855         if (!cfq_cfqq_sync(cfqq))
 856                 return ASYNC_WORKLOAD;
 857         if (!cfq_cfqq_idle_window(cfqq))
 858                 return SYNC_NOIDLE_WORKLOAD;
 859         return SYNC_WORKLOAD;
 860 }
 861
 862 static inline int cfq_group_busy_queues_wl(enum wl_class_t wl_class,
 863                                         struct cfq_data *cfqd,
 864                                         struct cfq_group *cfqg)
 865 {
 866         if (wl_class == IDLE_WORKLOAD)
 867                 return cfqg->service_tree_idle.count;
 868
 869         return cfqg->service_trees[wl_class][ASYNC_WORKLOAD].count +
 870                 cfqg->service_trees[wl_class][SYNC_NOIDLE_WORKLOAD].count +
 871                 cfqg->service_trees[wl_class][SYNC_WORKLOAD].count;
 872 }
 873
 874 static inline int cfqg_busy_async_queues(struct cfq_data *cfqd,
 875                                         struct cfq_group *cfqg)
 876 {
 877         return cfqg->service_trees[RT_WORKLOAD][ASYNC_WORKLOAD].count +
 878                 cfqg->service_trees[BE_WORKLOAD][ASYNC_WORKLOAD].count;
 879 }
 880
 881 static void cfq_dispatch_insert(struct request_queue *, struct request *);
 882 static struct cfq_queue *cfq_get_queue(struct cfq_data *cfqd, bool is_sync,
 883                                        struct cfq_io_cq *cic, struct bio *bio);
 884
 885 static inline struct cfq_io_cq *icq_to_cic(struct io_cq *icq)
 886 {
 887         /* cic->icq is the first member, %NULL will convert to %NULL */
 888         return container_of(icq, struct cfq_io_cq, icq);
 889 }
 890
 891 static inline struct cfq_io_cq *cfq_cic_lookup(struct cfq_data *cfqd,
 892                                                struct io_context *ioc)
 893 {
 894         if (ioc)
 895                 return icq_to_cic(ioc_lookup_icq(ioc, cfqd->queue));
 896         return NULL;
 897 }
 898
 899 static inline struct cfq_queue *cic_to_cfqq(struct cfq_io_cq *cic, bool is_sync)
 900 {
 901         return cic->cfqq[is_sync];
 902 }
 903
 904 static inline void cic_set_cfqq(struct cfq_io_cq *cic, struct cfq_queue *cfqq,
 905                                 bool is_sync)
 906 {
 907         cic->cfqq[is_sync] = cfqq;
 908 }
 909
 910 static inline struct cfq_data *cic_to_cfqd(struct cfq_io_cq *cic)
 911 {
 912         return cic->icq.q->elevator->elevator_data;
 913 }
 914
 915 /*
 916  * scheduler run of queue, if there are requests pending and no one in the
 917  * driver that will restart queueing
 918  */
 919 static inline void cfq_schedule_dispatch(struct cfq_data *cfqd)
 920 {
 921         if (cfqd->busy_queues) {
 922                 cfq_log(cfqd, "schedule dispatch");
 923                 kblockd_schedule_work(&cfqd->unplug_work);
 924         }
 925 }
 926
 927 /*
 928  * Scale schedule slice based on io priority. Use the sync time slice only
 929  * if a queue is marked sync and has sync io queued. A sync queue with async
 930  * io only, should not get full sync slice length.
 931  */
 932 static inline u64 cfq_prio_slice(struct cfq_data *cfqd, bool sync,
 933                                  unsigned short prio)
 934 {
 935         u64 base_slice = cfqd->cfq_slice[sync];
 936         u64 slice = div_u64(base_slice, CFQ_SLICE_SCALE);
 937
 938         WARN_ON(prio >= IOPRIO_BE_NR);
 939
 940         return base_slice + (slice * (4 - prio));
 941 }
 942
 943 static inline u64
 944 cfq_prio_to_slice(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 945 {
 946         return cfq_prio_slice(cfqd, cfq_cfqq_sync(cfqq), cfqq->ioprio);
 947 }
 948
 949 /**
 950  * cfqg_scale_charge - scale disk time charge according to cfqg weight
 951  * @charge: disk time being charged
 952  * @vfraction: vfraction of the cfqg, fixed point w/ CFQ_SERVICE_SHIFT
 953  *
 954  * Scale @charge according to @vfraction, which is in range (0, 1].  The
 955  * scaling is inversely proportional.
 956  *
 957  * scaled = charge / vfraction
 958  *
 959  * The result is also in fixed point w/ CFQ_SERVICE_SHIFT.
 960  */
 961 static inline u64 cfqg_scale_charge(u64 charge,
 962                                     unsigned int vfraction)
 963 {
 964         u64 c = charge << CFQ_SERVICE_SHIFT;    /* make it fixed point */
 965
 966         /* charge / vfraction */
 967         c <<= CFQ_SERVICE_SHIFT;
 968         return div_u64(c, vfraction);
 969 }
 970
 971 static inline u64 max_vdisktime(u64 min_vdisktime, u64 vdisktime)
 972 {
 973         s64 delta = (s64)(vdisktime - min_vdisktime);
 974         if (delta > 0)
 975                 min_vdisktime = vdisktime;
 976
 977         return min_vdisktime;
 978 }
 979
 980 static inline u64 min_vdisktime(u64 min_vdisktime, u64 vdisktime)
 981 {
 982         s64 delta = (s64)(vdisktime - min_vdisktime);
 983         if (delta < 0)
 984                 min_vdisktime = vdisktime;
 985
 986         return min_vdisktime;
 987 }
 988
 989 static void update_min_vdisktime(struct cfq_rb_root *st)
 990 {
 991         struct cfq_group *cfqg;
 992
 993         if (st->left) {
 994                 cfqg = rb_entry_cfqg(st->left);
 995                 st->min_vdisktime = max_vdisktime(st->min_vdisktime,
 996                                                   cfqg->vdisktime);
 997         }
 998 }
 999
1000 /*
1001  * get averaged number of queues of RT/BE priority.
1002  * average is updated, with a formula that gives more weight to higher numbers,
1003  * to quickly follows sudden increases and decrease slowly
1004  */
1005
1006 static inline unsigned cfq_group_get_avg_queues(struct cfq_data *cfqd,
1007                                         struct cfq_group *cfqg, bool rt)
1008 {
1009         unsigned min_q, max_q;
1010         unsigned mult  = cfq_hist_divisor - 1;
1011         unsigned round = cfq_hist_divisor / 2;
1012         unsigned busy = cfq_group_busy_queues_wl(rt, cfqd, cfqg);
1013
1014         min_q = min(cfqg->busy_queues_avg[rt], busy);
1015         max_q = max(cfqg->busy_queues_avg[rt], busy);
1016         cfqg->busy_queues_avg[rt] = (mult * max_q + min_q + round) /
1017                 cfq_hist_divisor;
1018         return cfqg->busy_queues_avg[rt];
1019 }
1020
1021 static inline u64
1022 cfq_group_slice(struct cfq_data *cfqd, struct cfq_group *cfqg)
1023 {
1024         return cfqd->cfq_target_latency * cfqg->vfraction >> CFQ_SERVICE_SHIFT;
1025 }
1026
1027 static inline u64
1028 cfq_scaled_cfqq_slice(struct cfq_data *cfqd, struct cfq_queue *cfqq)
1029 {
1030         u64 slice = cfq_prio_to_slice(cfqd, cfqq);
1031         if (cfqd->cfq_latency) {
1032                 /*
1033                  * interested queues (we consider only the ones with the same
1034                  * priority class in the cfq group)
1035                  */
1036                 unsigned iq = cfq_group_get_avg_queues(cfqd, cfqq->cfqg,
1037                                                 cfq_class_rt(cfqq));
1038                 u64 sync_slice = cfqd->cfq_slice[1];
1039                 u64 expect_latency = sync_slice * iq;
1040                 u64 group_slice = cfq_group_slice(cfqd, cfqq->cfqg);
1041
1042                 if (expect_latency > group_slice) {
1043                         u64 base_low_slice = 2 * cfqd->cfq_slice_idle;
1044                         u64 low_slice;
1045
1046                         /* scale low_slice according to IO priority
1047                          * and sync vs async */
1048                         low_slice = div64_u64(base_low_slice*slice, sync_slice);
1049                         low_slice = min(slice, low_slice);
1050                         /* the adapted slice value is scaled to fit all iqs
1051                          * into the target latency */
1052                         slice = div64_u64(slice*group_slice, expect_latency);
1053                         slice = max(slice, low_slice);
1054                 }
1055         }
1056         return slice;
1057 }
1058
1059 static inline void
1060 cfq_set_prio_slice(struct cfq_data *cfqd, struct cfq_queue *cfqq)
1061 {
1062         u64 slice = cfq_scaled_cfqq_slice(cfqd, cfqq);
1063         u64 now = ktime_get_ns();
1064
1065         cfqq->slice_start = now;
1066         cfqq->slice_end = now + slice;
1067         cfqq->allocated_slice = slice;
1068         cfq_log_cfqq(cfqd, cfqq, "set_slice=%llu", cfqq->slice_end - now);
1069 }
1070
1071 /*
1072  * We need to wrap this check in cfq_cfqq_slice_new(), since ->slice_end
1073  * isn't valid until the first request from the dispatch is activated
1074  * and the slice time set.
1075  */
1076 static inline bool cfq_slice_used(struct cfq_queue *cfqq)
1077 {
1078         if (cfq_cfqq_slice_new(cfqq))
1079                 return false;
1080         if (ktime_get_ns() < cfqq->slice_end)
1081                 return false;
1082
1083         return true;
1084 }
1085
1086 /*
1087  * Lifted from AS - choose which of rq1 and rq2 that is best served now.
1088  * We choose the request that is closest to the head right now. Distance
1089  * behind the head is penalized and only allowed to a certain extent.
1090  */
1091 static struct request *
1092 cfq_choose_req(struct cfq_data *cfqd, struct request *rq1, struct request *rq2, sector_t last)
1093 {
1094         sector_t s1, s2, d1 = 0, d2 = 0;
1095         unsigned long back_max;
1096 #define CFQ_RQ1_WRAP    0x01 /* request 1 wraps */
1097 #define CFQ_RQ2_WRAP    0x02 /* request 2 wraps */
1098         unsigned wrap = 0; /* bit mask: requests behind the disk head? */
1099
1100         if (rq1 == NULL || rq1 == rq2)
1101                 return rq2;
1102         if (rq2 == NULL)
1103                 return rq1;
1104
1105         if (rq_is_sync(rq1) != rq_is_sync(rq2))
1106                 return rq_is_sync(rq1) ? rq1 : rq2;
1107
1108         if ((rq1->cmd_flags ^ rq2->cmd_flags) & REQ_PRIO)
1109                 return rq1->cmd_flags & REQ_PRIO ? rq1 : rq2;
1110
1111         s1 = blk_rq_pos(rq1);
1112         s2 = blk_rq_pos(rq2);
1113
1114         /*
1115          * by definition, 1KiB is 2 sectors
1116          */
1117         back_max = cfqd->cfq_back_max * 2;
1118
1119         /*
1120          * Strict one way elevator _except_ in the case where we allow
1121          * short backward seeks which are biased as twice the cost of a
1122          * similar forward seek.
1123          */
1124         if (s1 >= last)
1125                 d1 = s1 - last;
1126         else if (s1 + back_max >= last)
1127                 d1 = (last - s1) * cfqd->cfq_back_penalty;
1128         else
1129                 wrap |= CFQ_RQ1_WRAP;
1130
1131         if (s2 >= last)
1132                 d2 = s2 - last;
1133         else if (s2 + back_max >= last)
1134                 d2 = (last - s2) * cfqd->cfq_back_penalty;
1135         else
1136                 wrap |= CFQ_RQ2_WRAP;
1137
1138         /* Found required data */
1139
1140         /*
1141          * By doing switch() on the bit mask "wrap" we avoid having to
1142          * check two variables for all permutations: --> faster!
1143          */
1144         switch (wrap) {
1145         case 0: /* common case for CFQ: rq1 and rq2 not wrapped */
1146                 if (d1 < d2)
1147                         return rq1;
1148                 else if (d2 < d1)
1149                         return rq2;
1150                 else {
1151                         if (s1 >= s2)
1152                                 return rq1;
1153                         else
1154                                 return rq2;
1155                 }
1156
1157         case CFQ_RQ2_WRAP:
1158                 return rq1;
1159         case CFQ_RQ1_WRAP:
1160                 return rq2;
1161         case (CFQ_RQ1_WRAP|CFQ_RQ2_WRAP): /* both rqs wrapped */
1162         default:
1163                 /*
1164                  * Since both rqs are wrapped,
1165                  * start with the one that's further behind head
1166                  * (--> only *one* back seek required),
1167                  * since back seek takes more time than forward.
1168                  */
1169                 if (s1 <= s2)
1170                         return rq1;
1171                 else
1172                         return rq2;
1173         }
1174 }
1175
1176 /*
1177  * The below is leftmost cache rbtree addon
1178  */
1179 static struct cfq_queue *cfq_rb_first(struct cfq_rb_root *root)
1180 {
1181         /* Service tree is empty */
1182         if (!root->count)
1183                 return NULL;
1184
1185         if (!root->left)
1186                 root->left = rb_first(&root->rb);
1187
1188         if (root->left)
1189                 return rb_entry(root->left, struct cfq_queue, rb_node);
1190
1191         return NULL;
1192 }
1193
1194 static struct cfq_group *cfq_rb_first_group(struct cfq_rb_root *root)
1195 {
1196         if (!root->left)
1197                 root->left = rb_first(&root->rb);
1198
1199         if (root->left)
1200                 return rb_entry_cfqg(root->left);
1201
1202         return NULL;
1203 }
1204
1205 static void rb_erase_init(struct rb_node *n, struct rb_root *root)
1206 {
1207         rb_erase(n, root);
1208         RB_CLEAR_NODE(n);
1209 }
1210
1211 static void cfq_rb_erase(struct rb_node *n, struct cfq_rb_root *root)
1212 {
1213         if (root->left == n)
1214                 root->left = NULL;
1215         rb_erase_init(n, &root->rb);
1216         --root->count;
1217 }
1218
1219 /*
1220  * would be nice to take fifo expire time into account as well
1221  */
1222 static struct request *
1223 cfq_find_next_rq(struct cfq_data *cfqd, struct cfq_queue *cfqq,
1224                   struct request *last)
1225 {
1226         struct rb_node *rbnext = rb_next(&last->rb_node);
1227         struct rb_node *rbprev = rb_prev(&last->rb_node);
1228         struct request *next = NULL, *prev = NULL;
1229
1230         BUG_ON(RB_EMPTY_NODE(&last->rb_node));
1231
1232         if (rbprev)
1233                 prev = rb_entry_rq(rbprev);
1234
1235         if (rbnext)
1236                 next = rb_entry_rq(rbnext);
1237         else {
1238                 rbnext = rb_first(&cfqq->sort_list);
1239                 if (rbnext && rbnext != &last->rb_node)
1240                         next = rb_entry_rq(rbnext);
1241         }
1242
1243         return cfq_choose_req(cfqd, next, prev, blk_rq_pos(last));
1244 }
1245
1246 static u64 cfq_slice_offset(struct cfq_data *cfqd,
1247                             struct cfq_queue *cfqq)
1248 {
1249         /*
1250          * just an approximation, should be ok.
1251          */
1252         return (cfqq->cfqg->nr_cfqq - 1) * (cfq_prio_slice(cfqd, 1, 0) -
1253                        cfq_prio_slice(cfqd, cfq_cfqq_sync(cfqq), cfqq->ioprio));
1254 }
1255
1256 static inline s64
1257 cfqg_key(struct cfq_rb_root *st, struct cfq_group *cfqg)
1258 {
1259         return cfqg->vdisktime - st->min_vdisktime;
1260 }
1261
1262 static void
1263 __cfq_group_service_tree_add(struct cfq_rb_root *st, struct cfq_group *cfqg)
1264 {
1265         struct rb_node **node = &st->rb.rb_node;
1266         struct rb_node *parent = NULL;
1267         struct cfq_group *__cfqg;
1268         s64 key = cfqg_key(st, cfqg);
1269         int left = 1;
1270
1271         while (*node != NULL) {
1272                 parent = *node;
1273                 __cfqg = rb_entry_cfqg(parent);
1274
1275                 if (key < cfqg_key(st, __cfqg))
1276                         node = &parent->rb_left;
1277                 else {
1278                         node = &parent->rb_right;
1279                         left = 0;
1280                 }
1281         }
1282
1283         if (left)
1284                 st->left = &cfqg->rb_node;
1285
1286         rb_link_node(&cfqg->rb_node, parent, node);
1287         rb_insert_color(&cfqg->rb_node, &st->rb);
1288 }
1289
1290 /*
1291  * This has to be called only on activation of cfqg
1292  */
1293 static void
1294 cfq_update_group_weight(struct cfq_group *cfqg)
1295 {
1296         if (cfqg->new_weight) {
1297                 cfqg->weight = cfqg->new_weight;
1298                 cfqg->new_weight = 0;
1299         }
1300 }
1301
1302 static void
1303 cfq_update_group_leaf_weight(struct cfq_group *cfqg)
1304 {
1305         BUG_ON(!RB_EMPTY_NODE(&cfqg->rb_node));
1306
1307         if (cfqg->new_leaf_weight) {
1308                 cfqg->leaf_weight = cfqg->new_leaf_weight;
1309                 cfqg->new_leaf_weight = 0;
1310         }
1311 }
1312
1313 static void
1314 cfq_group_service_tree_add(struct cfq_rb_root *st, struct cfq_group *cfqg)
1315 {
1316         unsigned int vfr = 1 << CFQ_SERVICE_SHIFT;      /* start with 1 */
1317         struct cfq_group *pos = cfqg;
1318         struct cfq_group *parent;
1319         bool propagate;
1320
1321         /* add to the service tree */
1322         BUG_ON(!RB_EMPTY_NODE(&cfqg->rb_node));
1323
1324         /*
1325          * Update leaf_weight.  We cannot update weight at this point
1326          * because cfqg might already have been activated and is
1327          * contributing its current weight to the parent's child_weight.
1328          */
1329         cfq_update_group_leaf_weight(cfqg);
1330         __cfq_group_service_tree_add(st, cfqg);
1331
1332         /*
1333          * Activate @cfqg and calculate the portion of vfraction @cfqg is
1334          * entitled to.  vfraction is calculated by walking the tree
1335          * towards the root calculating the fraction it has at each level.
1336          * The compounded ratio is how much vfraction @cfqg owns.
1337          *
1338          * Start with the proportion tasks in this cfqg has against active
1339          * children cfqgs - its leaf_weight against children_weight.
1340          */
1341         propagate = !pos->nr_active++;
1342         pos->children_weight += pos->leaf_weight;
1343         vfr = vfr * pos->leaf_weight / pos->children_weight;
1344
1345         /*
1346          * Compound ->weight walking up the tree.  Both activation and
1347          * vfraction calculation are done in the same loop.  Propagation
1348          * stops once an already activated node is met.  vfraction
1349          * calculation should always continue to the root.
1350          */
1351         while ((parent = cfqg_parent(pos))) {
1352                 if (propagate) {
1353                         cfq_update_group_weight(pos);
1354                         propagate = !parent->nr_active++;
1355                         parent->children_weight += pos->weight;
1356                 }
1357                 vfr = vfr * pos->weight / parent->children_weight;
1358                 pos = parent;
1359         }
1360
1361         cfqg->vfraction = max_t(unsigned, vfr, 1);
1362 }
1363
1364 static void
1365 cfq_group_notify_queue_add(struct cfq_data *cfqd, struct cfq_group *cfqg)
1366 {
1367         struct cfq_rb_root *st = &cfqd->grp_service_tree;
1368         struct cfq_group *__cfqg;
1369         struct rb_node *n;
1370
1371         cfqg->nr_cfqq++;
1372         if (!RB_EMPTY_NODE(&cfqg->rb_node))
1373                 return;
1374
1375         /*
1376          * Currently put the group at the end. Later implement something
1377          * so that groups get lesser vtime based on their weights, so that
1378          * if group does not loose all if it was not continuously backlogged.
1379          */
1380         n = rb_last(&st->rb);
1381         if (n) {
1382                 __cfqg = rb_entry_cfqg(n);
1383                 cfqg->vdisktime = __cfqg->vdisktime + CFQ_IDLE_DELAY;
1384         } else
1385                 cfqg->vdisktime = st->min_vdisktime;
1386         cfq_group_service_tree_add(st, cfqg);
1387 }
1388
1389 static void
1390 cfq_group_service_tree_del(struct cfq_rb_root *st, struct cfq_group *cfqg)
1391 {
1392         struct cfq_group *pos = cfqg;
1393         bool propagate;
1394
1395         /*
1396          * Undo activation from cfq_group_service_tree_add().  Deactivate
1397          * @cfqg and propagate deactivation upwards.
1398          */
1399         propagate = !--pos->nr_active;
1400         pos->children_weight -= pos->leaf_weight;
1401
1402         while (propagate) {
1403                 struct cfq_group *parent = cfqg_parent(pos);
1404
1405                 /* @pos has 0 nr_active at this point */
1406                 WARN_ON_ONCE(pos->children_weight);
1407                 pos->vfraction = 0;
1408
1409                 if (!parent)
1410                         break;
1411
1412                 propagate = !--parent->nr_active;
1413                 parent->children_weight -= pos->weight;
1414                 pos = parent;
1415         }
1416
1417         /* remove from the service tree */
1418         if (!RB_EMPTY_NODE(&cfqg->rb_node))
1419                 cfq_rb_erase(&cfqg->rb_node, st);
1420 }
1421
1422 static void
1423 cfq_group_notify_queue_del(struct cfq_data *cfqd, struct cfq_group *cfqg)
1424 {
1425         struct cfq_rb_root *st = &cfqd->grp_service_tree;
1426
1427         BUG_ON(cfqg->nr_cfqq < 1);
1428         cfqg->nr_cfqq--;
1429
1430         /* If there are other cfq queues under this group, don't delete it */
1431         if (cfqg->nr_cfqq)
1432                 return;
1433
1434         cfq_log_cfqg(cfqd, cfqg, "del_from_rr group");
1435         cfq_group_service_tree_del(st, cfqg);
1436         cfqg->saved_wl_slice = 0;
1437         cfqg_stats_update_dequeue(cfqg);
1438 }
1439
1440 static inline u64 cfq_cfqq_slice_usage(struct cfq_queue *cfqq,
1441                                        u64 *unaccounted_time)
1442 {
1443         u64 slice_used;
1444         u64 now = ktime_get_ns();
1445
1446         /*
1447          * Queue got expired before even a single request completed or
1448          * got expired immediately after first request completion.
1449          */
1450         if (!cfqq->slice_start || cfqq->slice_start == now) {
1451                 /*
1452                  * Also charge the seek time incurred to the group, otherwise
1453                  * if there are mutiple queues in the group, each can dispatch
1454                  * a single request on seeky media and cause lots of seek time
1455                  * and group will never know it.
1456                  */
1457                 slice_used = max_t(u64, (now - cfqq->dispatch_start),
1458                                         jiffies_to_nsecs(1));
1459         } else {
1460                 slice_used = now - cfqq->slice_start;
1461                 if (slice_used > cfqq->allocated_slice) {
1462                         *unaccounted_time = slice_used - cfqq->allocated_slice;
1463                         slice_used = cfqq->allocated_slice;
1464                 }
1465                 if (cfqq->slice_start > cfqq->dispatch_start)
1466                         *unaccounted_time += cfqq->slice_start -
1467                                         cfqq->dispatch_start;
1468         }
1469
1470         return slice_used;
1471 }
1472
1473 static void cfq_group_served(struct cfq_data *cfqd, struct cfq_group *cfqg,
1474                                 struct cfq_queue *cfqq)
1475 {
1476         struct cfq_rb_root *st = &cfqd->grp_service_tree;
1477         u64 used_sl, charge, unaccounted_sl = 0;
1478         int nr_sync = cfqg->nr_cfqq - cfqg_busy_async_queues(cfqd, cfqg)
1479                         - cfqg->service_tree_idle.count;
1480         unsigned int vfr;
1481         u64 now = ktime_get_ns();
1482
1483         BUG_ON(nr_sync < 0);
1484         used_sl = charge = cfq_cfqq_slice_usage(cfqq, &unaccounted_sl);
1485
1486         if (iops_mode(cfqd))
1487                 charge = cfqq->slice_dispatch;
1488         else if (!cfq_cfqq_sync(cfqq) && !nr_sync)
1489                 charge = cfqq->allocated_slice;
1490
1491         /*
1492          * Can't update vdisktime while on service tree and cfqg->vfraction
1493          * is valid only while on it.  Cache vfr, leave the service tree,
1494          * update vdisktime and go back on.  The re-addition to the tree
1495          * will also update the weights as necessary.
1496          */
1497         vfr = cfqg->vfraction;
1498         cfq_group_service_tree_del(st, cfqg);
1499         cfqg->vdisktime += cfqg_scale_charge(charge, vfr);
1500         cfq_group_service_tree_add(st, cfqg);
1501
1502         /* This group is being expired. Save the context */
1503         if (cfqd->workload_expires > now) {
1504                 cfqg->saved_wl_slice = cfqd->workload_expires - now;
1505                 cfqg->saved_wl_type = cfqd->serving_wl_type;
1506                 cfqg->saved_wl_class = cfqd->serving_wl_class;
1507         } else
1508                 cfqg->saved_wl_slice = 0;
1509
1510         cfq_log_cfqg(cfqd, cfqg, "served: vt=%llu min_vt=%llu", cfqg->vdisktime,
1511                                         st->min_vdisktime);
1512         cfq_log_cfqq(cfqq->cfqd, cfqq,
1513                      "sl_used=%llu disp=%llu charge=%llu iops=%u sect=%lu",
1514                      used_sl, cfqq->slice_dispatch, charge,
1515                      iops_mode(cfqd), cfqq->nr_sectors);
1516         cfqg_stats_update_timeslice_used(cfqg, used_sl, unaccounted_sl);
1517         cfqg_stats_set_start_empty_time(cfqg);
1518 }
1519
1520 /**
1521  * cfq_init_cfqg_base - initialize base part of a cfq_group
1522  * @cfqg: cfq_group to initialize
1523  *
1524  * Initialize the base part which is used whether %CONFIG_CFQ_GROUP_IOSCHED
1525  * is enabled or not.
1526  */
1527 static void cfq_init_cfqg_base(struct cfq_group *cfqg)
1528 {
1529         struct cfq_rb_root *st;
1530         int i, j;
1531
1532         for_each_cfqg_st(cfqg, i, j, st)
1533                 *st = CFQ_RB_ROOT;
1534         RB_CLEAR_NODE(&cfqg->rb_node);
1535
1536         cfqg->ttime.last_end_request = ktime_get_ns();
1537 }
1538
1539 #ifdef CONFIG_CFQ_GROUP_IOSCHED
1540 static int __cfq_set_weight(struct cgroup_subsys_state *css, u64 val,
1541                             bool on_dfl, bool reset_dev, bool is_leaf_weight);
1542
1543 static void cfqg_stats_exit(struct cfqg_stats *stats)
1544 {
1545         blkg_rwstat_exit(&stats->merged);
1546         blkg_rwstat_exit(&stats->service_time);
1547         blkg_rwstat_exit(&stats->wait_time);
1548         blkg_rwstat_exit(&stats->queued);
1549         blkg_stat_exit(&stats->time);
1550 #ifdef CONFIG_DEBUG_BLK_CGROUP
1551         blkg_stat_exit(&stats->unaccounted_time);
1552         blkg_stat_exit(&stats->avg_queue_size_sum);
1553         blkg_stat_exit(&stats->avg_queue_size_samples);
1554         blkg_stat_exit(&stats->dequeue);
1555         blkg_stat_exit(&stats->group_wait_time);
1556         blkg_stat_exit(&stats->idle_time);
1557         blkg_stat_exit(&stats->empty_time);
1558 #endif
1559 }
1560
1561 static int cfqg_stats_init(struct cfqg_stats *stats, gfp_t gfp)
1562 {
1563         if (blkg_rwstat_init(&stats->merged, gfp) ||
1564             blkg_rwstat_init(&stats->service_time, gfp) ||
1565             blkg_rwstat_init(&stats->wait_time, gfp) ||
1566             blkg_rwstat_init(&stats->queued, gfp) ||
1567             blkg_stat_init(&stats->time, gfp))
1568                 goto err;
1569
1570 #ifdef CONFIG_DEBUG_BLK_CGROUP
1571         if (blkg_stat_init(&stats->unaccounted_time, gfp) ||
1572             blkg_stat_init(&stats->avg_queue_size_sum, gfp) ||
1573             blkg_stat_init(&stats->avg_queue_size_samples, gfp) ||
1574             blkg_stat_init(&stats->dequeue, gfp) ||
1575             blkg_stat_init(&stats->group_wait_time, gfp) ||
1576             blkg_stat_init(&stats->idle_time, gfp) ||
1577             blkg_stat_init(&stats->empty_time, gfp))
1578                 goto err;
1579 #endif
1580         return 0;
1581 err:
1582         cfqg_stats_exit(stats);
1583         return -ENOMEM;
1584 }
1585
1586 static struct blkcg_policy_data *cfq_cpd_alloc(gfp_t gfp)
1587 {
1588         struct cfq_group_data *cgd;
1589
1590         cgd = kzalloc(sizeof(*cgd), gfp);
1591         if (!cgd)
1592                 return NULL;
1593         return &cgd->cpd;
1594 }
1595
1596 static void cfq_cpd_init(struct blkcg_policy_data *cpd)
1597 {
1598         struct cfq_group_data *cgd = cpd_to_cfqgd(cpd);
1599         unsigned int weight = cgroup_subsys_on_dfl(io_cgrp_subsys) ?
1600                               CGROUP_WEIGHT_DFL : CFQ_WEIGHT_LEGACY_DFL;
1601
1602         if (cpd_to_blkcg(cpd) == &blkcg_root)
1603                 weight *= 2;
1604
1605         cgd->weight = weight;
1606         cgd->leaf_weight = weight;
1607 }
1608
1609 static void cfq_cpd_free(struct blkcg_policy_data *cpd)
1610 {
1611         kfree(cpd_to_cfqgd(cpd));
1612 }
1613
1614 static void cfq_cpd_bind(struct blkcg_policy_data *cpd)
1615 {
1616         struct blkcg *blkcg = cpd_to_blkcg(cpd);
1617         bool on_dfl = cgroup_subsys_on_dfl(io_cgrp_subsys);
1618         unsigned int weight = on_dfl ? CGROUP_WEIGHT_DFL : CFQ_WEIGHT_LEGACY_DFL;
1619
1620         if (blkcg == &blkcg_root)
1621                 weight *= 2;
1622
1623         WARN_ON_ONCE(__cfq_set_weight(&blkcg->css, weight, on_dfl, true, false));
1624         WARN_ON_ONCE(__cfq_set_weight(&blkcg->css, weight, on_dfl, true, true));
1625 }
1626
1627 static struct blkg_policy_data *cfq_pd_alloc(gfp_t gfp, int node)
1628 {
1629         struct cfq_group *cfqg;
1630
1631         cfqg = kzalloc_node(sizeof(*cfqg), gfp, node);
1632         if (!cfqg)
1633                 return NULL;
1634
1635         cfq_init_cfqg_base(cfqg);
1636         if (cfqg_stats_init(&cfqg->stats, gfp)) {
1637                 kfree(cfqg);
1638                 return NULL;
1639         }
1640
1641         return &cfqg->pd;
1642 }
1643
1644 static void cfq_pd_init(struct blkg_policy_data *pd)
1645 {
1646         struct cfq_group *cfqg = pd_to_cfqg(pd);
1647         struct cfq_group_data *cgd = blkcg_to_cfqgd(pd->blkg->blkcg);
1648
1649         cfqg->weight = cgd->weight;
1650         cfqg->leaf_weight = cgd->leaf_weight;
1651 }
1652
1653 static void cfq_pd_offline(struct blkg_policy_data *pd)
1654 {
1655         struct cfq_group *cfqg = pd_to_cfqg(pd);
1656         int i;
1657
1658         for (i = 0; i < IOPRIO_BE_NR; i++) {
1659                 if (cfqg->async_cfqq[0][i])
1660                         cfq_put_queue(cfqg->async_cfqq[0][i]);
1661                 if (cfqg->async_cfqq[1][i])
1662                         cfq_put_queue(cfqg->async_cfqq[1][i]);
1663         }
1664
1665         if (cfqg->async_idle_cfqq)
1666                 cfq_put_queue(cfqg->async_idle_cfqq);
1667
1668         /*
1669          * @blkg is going offline and will be ignored by
1670          * blkg_[rw]stat_recursive_sum().  Transfer stats to the parent so
1671          * that they don't get lost.  If IOs complete after this point, the
1672          * stats for them will be lost.  Oh well...
1673          */
1674         cfqg_stats_xfer_dead(cfqg);
1675 }
1676
1677 static void cfq_pd_free(struct blkg_policy_data *pd)
1678 {
1679         struct cfq_group *cfqg = pd_to_cfqg(pd);
1680
1681         cfqg_stats_exit(&cfqg->stats);
1682         return kfree(cfqg);
1683 }
1684
1685 static void cfq_pd_reset_stats(struct blkg_policy_data *pd)
1686 {
1687         struct cfq_group *cfqg = pd_to_cfqg(pd);
1688
1689         cfqg_stats_reset(&cfqg->stats);
1690 }
1691
1692 static struct cfq_group *cfq_lookup_cfqg(struct cfq_data *cfqd,
1693                                          struct blkcg *blkcg)
1694 {
1695         struct blkcg_gq *blkg;
1696
1697         blkg = blkg_lookup(blkcg, cfqd->queue);
1698         if (likely(blkg))
1699                 return blkg_to_cfqg(blkg);
1700         return NULL;
1701 }
1702
1703 static void cfq_link_cfqq_cfqg(struct cfq_queue *cfqq, struct cfq_group *cfqg)
1704 {
1705         cfqq->cfqg = cfqg;
1706         /* cfqq reference on cfqg */
1707         cfqg_get(cfqg);
1708 }
1709
1710 static u64 cfqg_prfill_weight_device(struct seq_file *sf,
1711                                      struct blkg_policy_data *pd, int off)
1712 {
1713         struct cfq_group *cfqg = pd_to_cfqg(pd);
1714
1715         if (!cfqg->dev_weight)
1716                 return 0;
1717         return __blkg_prfill_u64(sf, pd, cfqg->dev_weight);
1718 }
1719
1720 static int cfqg_print_weight_device(struct seq_file *sf, void *v)
1721 {
1722         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1723                           cfqg_prfill_weight_device, &blkcg_policy_cfq,
1724                           0, false);
1725         return 0;
1726 }
1727
1728 static u64 cfqg_prfill_leaf_weight_device(struct seq_file *sf,
1729                                           struct blkg_policy_data *pd, int off)
1730 {
1731         struct cfq_group *cfqg = pd_to_cfqg(pd);
1732
1733         if (!cfqg->dev_leaf_weight)
1734                 return 0;
1735         return __blkg_prfill_u64(sf, pd, cfqg->dev_leaf_weight);
1736 }
1737
1738 static int cfqg_print_leaf_weight_device(struct seq_file *sf, void *v)
1739 {
1740         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1741                           cfqg_prfill_leaf_weight_device, &blkcg_policy_cfq,
1742                           0, false);
1743         return 0;
1744 }
1745
1746 static int cfq_print_weight(struct seq_file *sf, void *v)
1747 {
1748         struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
1749         struct cfq_group_data *cgd = blkcg_to_cfqgd(blkcg);
1750         unsigned int val = 0;
1751
1752         if (cgd)
1753                 val = cgd->weight;
1754
1755         seq_printf(sf, "%u\n", val);
1756         return 0;
1757 }
1758
1759 static int cfq_print_leaf_weight(struct seq_file *sf, void *v)
1760 {
1761         struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
1762         struct cfq_group_data *cgd = blkcg_to_cfqgd(blkcg);
1763         unsigned int val = 0;
1764
1765         if (cgd)
1766                 val = cgd->leaf_weight;
1767
1768         seq_printf(sf, "%u\n", val);
1769         return 0;
1770 }
1771
1772 static ssize_t __cfqg_set_weight_device(struct kernfs_open_file *of,
1773                                         char *buf, size_t nbytes, loff_t off,
1774                                         bool on_dfl, bool is_leaf_weight)
1775 {
1776         unsigned int min = on_dfl ? CGROUP_WEIGHT_MIN : CFQ_WEIGHT_LEGACY_MIN;
1777         unsigned int max = on_dfl ? CGROUP_WEIGHT_MAX : CFQ_WEIGHT_LEGACY_MAX;
1778         struct blkcg *blkcg = css_to_blkcg(of_css(of));
1779         struct blkg_conf_ctx ctx;
1780         struct cfq_group *cfqg;
1781         struct cfq_group_data *cfqgd;
1782         int ret;
1783         u64 v;
1784
1785         ret = blkg_conf_prep(blkcg, &blkcg_policy_cfq, buf, &ctx);
1786         if (ret)
1787                 return ret;
1788
1789         if (sscanf(ctx.body, "%llu", &v) == 1) {
1790                 /* require "default" on dfl */
1791                 ret = -ERANGE;
1792                 if (!v && on_dfl)
1793                         goto out_finish;
1794         } else if (!strcmp(strim(ctx.body), "default")) {
1795                 v = 0;
1796         } else {
1797                 ret = -EINVAL;
1798                 goto out_finish;
1799         }
1800
1801         cfqg = blkg_to_cfqg(ctx.blkg);
1802         cfqgd = blkcg_to_cfqgd(blkcg);
1803
1804         ret = -ERANGE;
1805         if (!v || (v >= min && v <= max)) {
1806                 if (!is_leaf_weight) {
1807                         cfqg->dev_weight = v;
1808                         cfqg->new_weight = v ?: cfqgd->weight;
1809                 } else {
1810                         cfqg->dev_leaf_weight = v;
1811                         cfqg->new_leaf_weight = v ?: cfqgd->leaf_weight;
1812                 }
1813                 ret = 0;
1814         }
1815 out_finish:
1816         blkg_conf_finish(&ctx);
1817         return ret ?: nbytes;
1818 }
1819
1820 static ssize_t cfqg_set_weight_device(struct kernfs_open_file *of,
1821                                       char *buf, size_t nbytes, loff_t off)
1822 {
1823         return __cfqg_set_weight_device(of, buf, nbytes, off, false, false);
1824 }
1825
1826 static ssize_t cfqg_set_leaf_weight_device(struct kernfs_open_file *of,
1827                                            char *buf, size_t nbytes, loff_t off)
1828 {
1829         return __cfqg_set_weight_device(of, buf, nbytes, off, false, true);
1830 }
1831
1832 static int __cfq_set_weight(struct cgroup_subsys_state *css, u64 val,
1833                             bool on_dfl, bool reset_dev, bool is_leaf_weight)
1834 {
1835         unsigned int min = on_dfl ? CGROUP_WEIGHT_MIN : CFQ_WEIGHT_LEGACY_MIN;
1836         unsigned int max = on_dfl ? CGROUP_WEIGHT_MAX : CFQ_WEIGHT_LEGACY_MAX;
1837         struct blkcg *blkcg = css_to_blkcg(css);
1838         struct blkcg_gq *blkg;
1839         struct cfq_group_data *cfqgd;
1840         int ret = 0;
1841
1842         if (val < min || val > max)
1843                 return -ERANGE;
1844
1845         spin_lock_irq(&blkcg->lock);
1846         cfqgd = blkcg_to_cfqgd(blkcg);
1847         if (!cfqgd) {
1848                 ret = -EINVAL;
1849                 goto out;
1850         }
1851
1852         if (!is_leaf_weight)
1853                 cfqgd->weight = val;
1854         else
1855                 cfqgd->leaf_weight = val;
1856
1857         hlist_for_each_entry(blkg, &blkcg->blkg_list, blkcg_node) {
1858                 struct cfq_group *cfqg = blkg_to_cfqg(blkg);
1859
1860                 if (!cfqg)
1861                         continue;
1862
1863                 if (!is_leaf_weight) {
1864                         if (reset_dev)
1865                                 cfqg->dev_weight = 0;
1866                         if (!cfqg->dev_weight)
1867                                 cfqg->new_weight = cfqgd->weight;
1868                 } else {
1869                         if (reset_dev)
1870                                 cfqg->dev_leaf_weight = 0;
1871                         if (!cfqg->dev_leaf_weight)
1872                                 cfqg->new_leaf_weight = cfqgd->leaf_weight;
1873                 }
1874         }
1875
1876 out:
1877         spin_unlock_irq(&blkcg->lock);
1878         return ret;
1879 }
1880
1881 static int cfq_set_weight(struct cgroup_subsys_state *css, struct cftype *cft,
1882                           u64 val)
1883 {
1884         return __cfq_set_weight(css, val, false, false, false);
1885 }
1886
1887 static int cfq_set_leaf_weight(struct cgroup_subsys_state *css,
1888                                struct cftype *cft, u64 val)
1889 {
1890         return __cfq_set_weight(css, val, false, false, true);
1891 }
1892
1893 static int cfqg_print_stat(struct seq_file *sf, void *v)
1894 {
1895         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_stat,
1896                           &blkcg_policy_cfq, seq_cft(sf)->private, false);
1897         return 0;
1898 }
1899
1900 static int cfqg_print_rwstat(struct seq_file *sf, void *v)
1901 {
1902         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), blkg_prfill_rwstat,
1903                           &blkcg_policy_cfq, seq_cft(sf)->private, true);
1904         return 0;
1905 }
1906
1907 static u64 cfqg_prfill_stat_recursive(struct seq_file *sf,
1908                                       struct blkg_policy_data *pd, int off)
1909 {
1910         u64 sum = blkg_stat_recursive_sum(pd_to_blkg(pd),
1911                                           &blkcg_policy_cfq, off);
1912         return __blkg_prfill_u64(sf, pd, sum);
1913 }
1914
1915 static u64 cfqg_prfill_rwstat_recursive(struct seq_file *sf,
1916                                         struct blkg_policy_data *pd, int off)
1917 {
1918         struct blkg_rwstat sum = blkg_rwstat_recursive_sum(pd_to_blkg(pd),
1919                                                         &blkcg_policy_cfq, off);
1920         return __blkg_prfill_rwstat(sf, pd, &sum);
1921 }
1922
1923 static int cfqg_print_stat_recursive(struct seq_file *sf, void *v)
1924 {
1925         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1926                           cfqg_prfill_stat_recursive, &blkcg_policy_cfq,
1927                           seq_cft(sf)->private, false);
1928         return 0;
1929 }
1930
1931 static int cfqg_print_rwstat_recursive(struct seq_file *sf, void *v)
1932 {
1933         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1934                           cfqg_prfill_rwstat_recursive, &blkcg_policy_cfq,
1935                           seq_cft(sf)->private, true);
1936         return 0;
1937 }
1938
1939 static u64 cfqg_prfill_sectors(struct seq_file *sf, struct blkg_policy_data *pd,
1940                                int off)
1941 {
1942         u64 sum = blkg_rwstat_total(&pd->blkg->stat_bytes);
1943
1944         return __blkg_prfill_u64(sf, pd, sum >> 9);
1945 }
1946
1947 static int cfqg_print_stat_sectors(struct seq_file *sf, void *v)
1948 {
1949         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1950                           cfqg_prfill_sectors, &blkcg_policy_cfq, 0, false);
1951         return 0;
1952 }
1953
1954 static u64 cfqg_prfill_sectors_recursive(struct seq_file *sf,
1955                                          struct blkg_policy_data *pd, int off)
1956 {
1957         struct blkg_rwstat tmp = blkg_rwstat_recursive_sum(pd->blkg, NULL,
1958                                         offsetof(struct blkcg_gq, stat_bytes));
1959         u64 sum = atomic64_read(&tmp.aux_cnt[BLKG_RWSTAT_READ]) +
1960                 atomic64_read(&tmp.aux_cnt[BLKG_RWSTAT_WRITE]);
1961
1962         return __blkg_prfill_u64(sf, pd, sum >> 9);
1963 }
1964
1965 static int cfqg_print_stat_sectors_recursive(struct seq_file *sf, void *v)
1966 {
1967         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1968                           cfqg_prfill_sectors_recursive, &blkcg_policy_cfq, 0,
1969                           false);
1970         return 0;
1971 }
1972
1973 #ifdef CONFIG_DEBUG_BLK_CGROUP
1974 static u64 cfqg_prfill_avg_queue_size(struct seq_file *sf,
1975                                       struct blkg_policy_data *pd, int off)
1976 {
1977         struct cfq_group *cfqg = pd_to_cfqg(pd);
1978         u64 samples = blkg_stat_read(&cfqg->stats.avg_queue_size_samples);
1979         u64 v = 0;
1980
1981         if (samples) {
1982                 v = blkg_stat_read(&cfqg->stats.avg_queue_size_sum);
1983                 v = div64_u64(v, samples);
1984         }
1985         __blkg_prfill_u64(sf, pd, v);
1986         return 0;
1987 }
1988
1989 /* print avg_queue_size */
1990 static int cfqg_print_avg_queue_size(struct seq_file *sf, void *v)
1991 {
1992         blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)),
1993                           cfqg_prfill_avg_queue_size, &blkcg_policy_cfq,
1994                           0, false);
1995         return 0;
1996 }
1997 #endif  /* CONFIG_DEBUG_BLK_CGROUP */
1998
1999 static struct cftype cfq_blkcg_legacy_files[] = {
2000         /* on root, weight is mapped to leaf_weight */
2001         {
2002                 .name = "weight_device",
2003                 .flags = CFTYPE_ONLY_ON_ROOT,
2004                 .seq_show = cfqg_print_leaf_weight_device,
2005                 .write = cfqg_set_leaf_weight_device,
2006         },
2007         {
2008                 .name = "weight",
2009                 .flags = CFTYPE_ONLY_ON_ROOT,
2010                 .seq_show = cfq_print_leaf_weight,
2011                 .write_u64 = cfq_set_leaf_weight,
2012         },
2013
2014         /* no such mapping necessary for !roots */
2015         {
2016                 .name = "weight_device",
2017                 .flags = CFTYPE_NOT_ON_ROOT,
2018                 .seq_show = cfqg_print_weight_device,
2019                 .write = cfqg_set_weight_device,
2020         },
2021         {
2022                 .name = "weight",
2023                 .flags = CFTYPE_NOT_ON_ROOT,
2024                 .seq_show = cfq_print_weight,
2025                 .write_u64 = cfq_set_weight,
2026         },
2027
2028         {
2029                 .name = "leaf_weight_device",
2030                 .seq_show = cfqg_print_leaf_weight_device,
2031                 .write = cfqg_set_leaf_weight_device,
2032         },
2033         {
2034                 .name = "leaf_weight",
2035                 .seq_show = cfq_print_leaf_weight,
2036                 .write_u64 = cfq_set_leaf_weight,
2037         },
2038
2039         /* statistics, covers only the tasks in the cfqg */
2040         {
2041                 .name = "time",
2042                 .private = offsetof(struct cfq_group, stats.time),
2043                 .seq_show = cfqg_print_stat,
2044         },
2045         {
2046                 .name = "sectors",
2047                 .seq_show = cfqg_print_stat_sectors,
2048         },
2049         {
2050                 .name = "io_service_bytes",
2051                 .private = (unsigned long)&blkcg_policy_cfq,
2052                 .seq_show = blkg_print_stat_bytes,
2053         },
2054         {
2055                 .name = "io_serviced",
2056                 .private = (unsigned long)&blkcg_policy_cfq,
2057                 .seq_show = blkg_print_stat_ios,
2058         },
2059         {
2060                 .name = "io_service_time",
2061                 .private = offsetof(struct cfq_group, stats.service_time),
2062                 .seq_show = cfqg_print_rwstat,
2063         },
2064         {
2065                 .name = "io_wait_time",
2066                 .private = offsetof(struct cfq_group, stats.wait_time),
2067                 .seq_show = cfqg_print_rwstat,
2068         },
2069         {
2070                 .name = "io_merged",
2071                 .private = offsetof(struct cfq_group, stats.merged),
2072                 .seq_show = cfqg_print_rwstat,
2073         },
2074         {
2075                 .name = "io_queued",
2076                 .private = offsetof(struct cfq_group, stats.queued),
2077                 .seq_show = cfqg_print_rwstat,
2078         },
2079
2080         /* the same statictics which cover the cfqg and its descendants */
2081         {
2082                 .name = "time_recursive",
2083                 .private = offsetof(struct cfq_group, stats.time),
2084                 .seq_show = cfqg_print_stat_recursive,
2085         },
2086         {
2087                 .name = "sectors_recursive",
2088                 .seq_show = cfqg_print_stat_sectors_recursive,
2089         },
2090         {
2091                 .name = "io_service_bytes_recursive",
2092                 .private = (unsigned long)&blkcg_policy_cfq,
2093                 .seq_show = blkg_print_stat_bytes_recursive,
2094         },
2095         {
2096                 .name = "io_serviced_recursive",
2097                 .private = (unsigned long)&blkcg_policy_cfq,
2098                 .seq_show = blkg_print_stat_ios_recursive,
2099         },
2100         {
2101                 .name = "io_service_time_recursive",
2102                 .private = offsetof(struct cfq_group, stats.service_time),
2103                 .seq_show = cfqg_print_rwstat_recursive,
2104         },
2105         {
2106                 .name = "io_wait_time_recursive",
2107                 .private = offsetof(struct cfq_group, stats.wait_time),
2108                 .seq_show = cfqg_print_rwstat_recursive,
2109         },
2110         {
2111                 .name = "io_merged_recursive",
2112                 .private = offsetof(struct cfq_group, stats.merged),
2113                 .seq_show = cfqg_print_rwstat_recursive,
2114         },
2115         {
2116                 .name = "io_queued_recursive",
2117                 .private = offsetof(struct cfq_group, stats.queued),
2118                 .seq_show = cfqg_print_rwstat_recursive,
2119         },
2120 #ifdef CONFIG_DEBUG_BLK_CGROUP
2121         {
2122                 .name = "avg_queue_size",
2123                 .seq_show = cfqg_print_avg_queue_size,
2124         },
2125         {
2126                 .name = "group_wait_time",
2127                 .private = offsetof(struct cfq_group, stats.group_wait_time),
2128                 .seq_show = cfqg_print_stat,
2129         },
2130         {
2131                 .name = "idle_time",
2132                 .private = offsetof(struct cfq_group, stats.idle_time),
2133                 .seq_show = cfqg_print_stat,
2134         },
2135         {
2136                 .name = "empty_time",
2137                 .private = offsetof(struct cfq_group, stats.empty_time),
2138                 .seq_show = cfqg_print_stat,
2139         },
2140         {
2141                 .name = "dequeue",
2142                 .private = offsetof(struct cfq_group, stats.dequeue),
2143                 .seq_show = cfqg_print_stat,
2144         },
2145         {
2146                 .name = "unaccounted_time",
2147                 .private = offsetof(struct cfq_group, stats.unaccounted_time),
2148                 .seq_show = cfqg_print_stat,
2149         },
2150 #endif  /* CONFIG_DEBUG_BLK_CGROUP */
2151         { }     /* terminate */
2152 };
2153
2154 static int cfq_print_weight_on_dfl(struct seq_file *sf, void *v)
2155 {
2156         struct blkcg *blkcg = css_to_blkcg(seq_css(sf));
2157         struct cfq_group_data *cgd = blkcg_to_cfqgd(blkcg);
2158
2159         seq_printf(sf, "default %u\n", cgd->weight);
2160         blkcg_print_blkgs(sf, blkcg, cfqg_prfill_weight_device,
2161                           &blkcg_policy_cfq, 0, false);
2162         return 0;
2163 }
2164
2165 static ssize_t cfq_set_weight_on_dfl(struct kernfs_open_file *of,
2166                                      char *buf, size_t nbytes, loff_t off)
2167 {
2168         char *endp;
2169         int ret;
2170         u64 v;
2171
2172         buf = strim(buf);
2173
2174         /* "WEIGHT" or "default WEIGHT" sets the default weight */
2175         v = simple_strtoull(buf, &endp, 0);
2176         if (*endp == '\0' || sscanf(buf, "default %llu", &v) == 1) {
2177                 ret = __cfq_set_weight(of_css(of), v, true, false, false);
2178                 return ret ?: nbytes;
2179         }
2180
2181         /* "MAJ:MIN WEIGHT" */
2182         return __cfqg_set_weight_device(of, buf, nbytes, off, true, false);
2183 }
2184
2185 static struct cftype cfq_blkcg_files[] = {
2186         {
2187                 .name = "weight",
2188                 .flags = CFTYPE_NOT_ON_ROOT,
2189                 .seq_show = cfq_print_weight_on_dfl,
2190                 .write = cfq_set_weight_on_dfl,
2191         },
2192         { }     /* terminate */
2193 };
2194
2195 #else /* GROUP_IOSCHED */
2196 static struct cfq_group *cfq_lookup_cfqg(struct cfq_data *cfqd,
2197                                          struct blkcg *blkcg)
2198 {
2199         return cfqd->root_group;
2200 }
2201
2202 static inline void
2203 cfq_link_cfqq_cfqg(struct cfq_queue *cfqq, struct cfq_group *cfqg) {
2204         cfqq->cfqg = cfqg;
2205 }
2206
2207 #endif /* GROUP_IOSCHED */
2208
2209 /*
2210  * The cfqd->service_trees holds all pending cfq_queue's that have
2211  * requests waiting to be processed. It is sorted in the order that
2212  * we will service the queues.
2213  */
2214 static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
2215                                  bool add_front)
2216 {
2217         struct rb_node **p, *parent;
2218         struct cfq_queue *__cfqq;
2219         u64 rb_key;
2220         struct cfq_rb_root *st;
2221         int left;
2222         int new_cfqq = 1;
2223         u64 now = ktime_get_ns();
2224
2225         st = st_for(cfqq->cfqg, cfqq_class(cfqq), cfqq_type(cfqq));
2226         if (cfq_class_idle(cfqq)) {
2227                 rb_key = CFQ_IDLE_DELAY;
2228                 parent = rb_last(&st->rb);
2229                 if (parent && parent != &cfqq->rb_node) {
2230                         __cfqq = rb_entry(parent, struct cfq_queue, rb_node);
2231                         rb_key += __cfqq->rb_key;
2232                 } else
2233                         rb_key += now;
2234         } else if (!add_front) {
2235                 /*
2236                  * Get our rb key offset. Subtract any residual slice
2237                  * value carried from last service. A negative resid
2238                  * count indicates slice overrun, and this should position
2239                  * the next service time further away in the tree.
2240                  */
2241                 rb_key = cfq_slice_offset(cfqd, cfqq) + now;
2242                 rb_key -= cfqq->slice_resid;
2243                 cfqq->slice_resid = 0;
2244         } else {
2245                 rb_key = -NSEC_PER_SEC;
2246                 __cfqq = cfq_rb_first(st);
2247                 rb_key += __cfqq ? __cfqq->rb_key : now;
2248         }
2249
2250         if (!RB_EMPTY_NODE(&cfqq->rb_node)) {
2251                 new_cfqq = 0;
2252                 /*
2253                  * same position, nothing more to do
2254                  */
2255                 if (rb_key == cfqq->rb_key && cfqq->service_tree == st)
2256                         return;
2257
2258                 cfq_rb_erase(&cfqq->rb_node, cfqq->service_tree);
2259                 cfqq->service_tree = NULL;
2260         }
2261
2262         left = 1;
2263         parent = NULL;
2264         cfqq->service_tree = st;
2265         p = &st->rb.rb_node;
2266         while (*p) {
2267                 parent = *p;
2268                 __cfqq = rb_entry(parent, struct cfq_queue, rb_node);
2269
2270                 /*
2271                  * sort by key, that represents service time.
2272                  */
2273                 if (rb_key < __cfqq->rb_key)
2274                         p = &parent->rb_left;
2275                 else {
2276                         p = &parent->rb_right;
2277                         left = 0;
2278                 }
2279         }
2280
2281         if (left)
2282                 st->left = &cfqq->rb_node;
2283
2284         cfqq->rb_key = rb_key;
2285         rb_link_node(&cfqq->rb_node, parent, p);
2286         rb_insert_color(&cfqq->rb_node, &st->rb);
2287         st->count++;
2288         if (add_front || !new_cfqq)
2289                 return;
2290         cfq_group_notify_queue_add(cfqd, cfqq->cfqg);
2291 }
2292
2293 static struct cfq_queue *
2294 cfq_prio_tree_lookup(struct cfq_data *cfqd, struct rb_root *root,
2295                      sector_t sector, struct rb_node **ret_parent,
2296                      struct rb_node ***rb_link)
2297 {
2298         struct rb_node **p, *parent;
2299         struct cfq_queue *cfqq = NULL;
2300
2301         parent = NULL;
2302         p = &root->rb_node;
2303         while (*p) {
2304                 struct rb_node **n;
2305
2306                 parent = *p;
2307                 cfqq = rb_entry(parent, struct cfq_queue, p_node);
2308
2309                 /*
2310                  * Sort strictly based on sector.  Smallest to the left,
2311                  * largest to the right.
2312                  */
2313                 if (sector > blk_rq_pos(cfqq->next_rq))
2314                         n = &(*p)->rb_right;
2315                 else if (sector < blk_rq_pos(cfqq->next_rq))
2316                         n = &(*p)->rb_left;
2317                 else
2318                         break;
2319                 p = n;
2320                 cfqq = NULL;
2321         }
2322
2323         *ret_parent = parent;
2324         if (rb_link)
2325                 *rb_link = p;
2326         return cfqq;
2327 }
2328
2329 static void cfq_prio_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2330 {
2331         struct rb_node **p, *parent;
2332         struct cfq_queue *__cfqq;
2333
2334         if (cfqq->p_root) {
2335                 rb_erase(&cfqq->p_node, cfqq->p_root);
2336                 cfqq->p_root = NULL;
2337         }
2338
2339         if (cfq_class_idle(cfqq))
2340                 return;
2341         if (!cfqq->next_rq)
2342                 return;
2343
2344         cfqq->p_root = &cfqd->prio_trees[cfqq->org_ioprio];
2345         __cfqq = cfq_prio_tree_lookup(cfqd, cfqq->p_root,
2346                                       blk_rq_pos(cfqq->next_rq), &parent, &p);
2347         if (!__cfqq) {
2348                 rb_link_node(&cfqq->p_node, parent, p);
2349                 rb_insert_color(&cfqq->p_node, cfqq->p_root);
2350         } else
2351                 cfqq->p_root = NULL;
2352 }
2353
2354 /*
2355  * Update cfqq's position in the service tree.
2356  */
2357 static void cfq_resort_rr_list(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2358 {
2359         /*
2360          * Resorting requires the cfqq to be on the RR list already.
2361          */
2362         if (cfq_cfqq_on_rr(cfqq)) {
2363                 cfq_service_tree_add(cfqd, cfqq, 0);
2364                 cfq_prio_tree_add(cfqd, cfqq);
2365         }
2366 }
2367
2368 /*
2369  * add to busy list of queues for service, trying to be fair in ordering
2370  * the pending list according to last request service
2371  */
2372 static void cfq_add_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2373 {
2374         cfq_log_cfqq(cfqd, cfqq, "add_to_rr");
2375         BUG_ON(cfq_cfqq_on_rr(cfqq));
2376         cfq_mark_cfqq_on_rr(cfqq);
2377         cfqd->busy_queues++;
2378         if (cfq_cfqq_sync(cfqq))
2379                 cfqd->busy_sync_queues++;
2380
2381         cfq_resort_rr_list(cfqd, cfqq);
2382 }
2383
2384 /*
2385  * Called when the cfqq no longer has requests pending, remove it from
2386  * the service tree.
2387  */
2388 static void cfq_del_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2389 {
2390         cfq_log_cfqq(cfqd, cfqq, "del_from_rr");
2391         BUG_ON(!cfq_cfqq_on_rr(cfqq));
2392         cfq_clear_cfqq_on_rr(cfqq);
2393
2394         if (!RB_EMPTY_NODE(&cfqq->rb_node)) {
2395                 cfq_rb_erase(&cfqq->rb_node, cfqq->service_tree);
2396                 cfqq->service_tree = NULL;
2397         }
2398         if (cfqq->p_root) {
2399                 rb_erase(&cfqq->p_node, cfqq->p_root);
2400                 cfqq->p_root = NULL;
2401         }
2402
2403         cfq_group_notify_queue_del(cfqd, cfqq->cfqg);
2404         BUG_ON(!cfqd->busy_queues);
2405         cfqd->busy_queues--;
2406         if (cfq_cfqq_sync(cfqq))
2407                 cfqd->busy_sync_queues--;
2408 }
2409
2410 /*
2411  * rb tree support functions
2412  */
2413 static void cfq_del_rq_rb(struct request *rq)
2414 {
2415         struct cfq_queue *cfqq = RQ_CFQQ(rq);
2416         const int sync = rq_is_sync(rq);
2417
2418         BUG_ON(!cfqq->queued[sync]);
2419         cfqq->queued[sync]--;
2420
2421         elv_rb_del(&cfqq->sort_list, rq);
2422
2423         if (cfq_cfqq_on_rr(cfqq) && RB_EMPTY_ROOT(&cfqq->sort_list)) {
2424                 /*
2425                  * Queue will be deleted from service tree when we actually
2426                  * expire it later. Right now just remove it from prio tree
2427                  * as it is empty.
2428                  */
2429                 if (cfqq->p_root) {
2430                         rb_erase(&cfqq->p_node, cfqq->p_root);
2431                         cfqq->p_root = NULL;
2432                 }
2433         }
2434 }
2435
2436 static void cfq_add_rq_rb(struct request *rq)
2437 {
2438         struct cfq_queue *cfqq = RQ_CFQQ(rq);
2439         struct cfq_data *cfqd = cfqq->cfqd;
2440         struct request *prev;
2441
2442         cfqq->queued[rq_is_sync(rq)]++;
2443
2444         elv_rb_add(&cfqq->sort_list, rq);
2445
2446         if (!cfq_cfqq_on_rr(cfqq))
2447                 cfq_add_cfqq_rr(cfqd, cfqq);
2448
2449         /*
2450          * check if this request is a better next-serve candidate
2451          */
2452         prev = cfqq->next_rq;
2453         cfqq->next_rq = cfq_choose_req(cfqd, cfqq->next_rq, rq, cfqd->last_position);
2454
2455         /*
2456          * adjust priority tree position, if ->next_rq changes
2457          */
2458         if (prev != cfqq->next_rq)
2459                 cfq_prio_tree_add(cfqd, cfqq);
2460
2461         BUG_ON(!cfqq->next_rq);
2462 }
2463
2464 static void cfq_reposition_rq_rb(struct cfq_queue *cfqq, struct request *rq)
2465 {
2466         elv_rb_del(&cfqq->sort_list, rq);
2467         cfqq->queued[rq_is_sync(rq)]--;
2468         cfqg_stats_update_io_remove(RQ_CFQG(rq), rq->cmd_flags);
2469         cfq_add_rq_rb(rq);
2470         cfqg_stats_update_io_add(RQ_CFQG(rq), cfqq->cfqd->serving_group,
2471                                  rq->cmd_flags);
2472 }
2473
2474 static struct request *
2475 cfq_find_rq_fmerge(struct cfq_data *cfqd, struct bio *bio)
2476 {
2477         struct task_struct *tsk = current;
2478         struct cfq_io_cq *cic;
2479         struct cfq_queue *cfqq;
2480
2481         cic = cfq_cic_lookup(cfqd, tsk->io_context);
2482         if (!cic)
2483                 return NULL;
2484
2485         cfqq = cic_to_cfqq(cic, op_is_sync(bio->bi_opf));
2486         if (cfqq)
2487                 return elv_rb_find(&cfqq->sort_list, bio_end_sector(bio));
2488
2489         return NULL;
2490 }
2491
2492 static void cfq_activate_request(struct request_queue *q, struct request *rq)
2493 {
2494         struct cfq_data *cfqd = q->elevator->elevator_data;
2495
2496         cfqd->rq_in_driver++;
2497         cfq_log_cfqq(cfqd, RQ_CFQQ(rq), "activate rq, drv=%d",
2498                                                 cfqd->rq_in_driver);
2499
2500         cfqd->last_position = blk_rq_pos(rq) + blk_rq_sectors(rq);
2501 }
2502
2503 static void cfq_deactivate_request(struct request_queue *q, struct request *rq)
2504 {
2505         struct cfq_data *cfqd = q->elevator->elevator_data;
2506
2507         WARN_ON(!cfqd->rq_in_driver);
2508         cfqd->rq_in_driver--;
2509         cfq_log_cfqq(cfqd, RQ_CFQQ(rq), "deactivate rq, drv=%d",
2510                                                 cfqd->rq_in_driver);
2511 }
2512
2513 static void cfq_remove_request(struct request *rq)
2514 {
2515         struct cfq_queue *cfqq = RQ_CFQQ(rq);
2516
2517         if (cfqq->next_rq == rq)
2518                 cfqq->next_rq = cfq_find_next_rq(cfqq->cfqd, cfqq, rq);
2519
2520         list_del_init(&rq->queuelist);
2521         cfq_del_rq_rb(rq);
2522
2523         cfqq->cfqd->rq_queued--;
2524         cfqg_stats_update_io_remove(RQ_CFQG(rq), rq->cmd_flags);
2525         if (rq->cmd_flags & REQ_PRIO) {
2526                 WARN_ON(!cfqq->prio_pending);
2527                 cfqq->prio_pending--;
2528         }
2529 }
2530
2531 static int cfq_merge(struct request_queue *q, struct request **req,
2532                      struct bio *bio)
2533 {
2534         struct cfq_data *cfqd = q->elevator->elevator_data;
2535         struct request *__rq;
2536
2537         __rq = cfq_find_rq_fmerge(cfqd, bio);
2538         if (__rq && elv_bio_merge_ok(__rq, bio)) {
2539                 *req = __rq;
2540                 return ELEVATOR_FRONT_MERGE;
2541         }
2542
2543         return ELEVATOR_NO_MERGE;
2544 }
2545
2546 static void cfq_merged_request(struct request_queue *q, struct request *req,
2547                                int type)
2548 {
2549         if (type == ELEVATOR_FRONT_MERGE) {
2550                 struct cfq_queue *cfqq = RQ_CFQQ(req);
2551
2552                 cfq_reposition_rq_rb(cfqq, req);
2553         }
2554 }
2555
2556 static void cfq_bio_merged(struct request_queue *q, struct request *req,
2557                                 struct bio *bio)
2558 {
2559         cfqg_stats_update_io_merged(RQ_CFQG(req), bio->bi_opf);
2560 }
2561
2562 static void
2563 cfq_merged_requests(struct request_queue *q, struct request *rq,
2564                     struct request *next)
2565 {
2566         struct cfq_queue *cfqq = RQ_CFQQ(rq);
2567         struct cfq_data *cfqd = q->elevator->elevator_data;
2568
2569         /*
2570          * reposition in fifo if next is older than rq
2571          */
2572         if (!list_empty(&rq->queuelist) && !list_empty(&next->queuelist) &&
2573             next->fifo_time < rq->fifo_time &&
2574             cfqq == RQ_CFQQ(next)) {
2575                 list_move(&rq->queuelist, &next->queuelist);
2576                 rq->fifo_time = next->fifo_time;
2577         }
2578
2579         if (cfqq->next_rq == next)
2580                 cfqq->next_rq = rq;
2581         cfq_remove_request(next);
2582         cfqg_stats_update_io_merged(RQ_CFQG(rq), next->cmd_flags);
2583
2584         cfqq = RQ_CFQQ(next);
2585         /*
2586          * all requests of this queue are merged to other queues, delete it
2587          * from the service tree. If it's the active_queue,
2588          * cfq_dispatch_requests() will choose to expire it or do idle
2589          */
2590         if (cfq_cfqq_on_rr(cfqq) && RB_EMPTY_ROOT(&cfqq->sort_list) &&
2591             cfqq != cfqd->active_queue)
2592                 cfq_del_cfqq_rr(cfqd, cfqq);
2593 }
2594
2595 static int cfq_allow_bio_merge(struct request_queue *q, struct request *rq,
2596                                struct bio *bio)
2597 {
2598         struct cfq_data *cfqd = q->elevator->elevator_data;
2599         bool is_sync = op_is_sync(bio->bi_opf);
2600         struct cfq_io_cq *cic;
2601         struct cfq_queue *cfqq;
2602
2603         /*
2604          * Disallow merge of a sync bio into an async request.
2605          */
2606         if (is_sync && !rq_is_sync(rq))
2607                 return false;
2608
2609         /*
2610          * Lookup the cfqq that this bio will be queued with and allow
2611          * merge only if rq is queued there.
2612          */
2613         cic = cfq_cic_lookup(cfqd, current->io_context);
2614         if (!cic)
2615                 return false;
2616
2617         cfqq = cic_to_cfqq(cic, is_sync);
2618         return cfqq == RQ_CFQQ(rq);
2619 }
2620
2621 static int cfq_allow_rq_merge(struct request_queue *q, struct request *rq,
2622                               struct request *next)
2623 {
2624         return RQ_CFQQ(rq) == RQ_CFQQ(next);
2625 }
2626
2627 static inline void cfq_del_timer(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2628 {
2629         hrtimer_try_to_cancel(&cfqd->idle_slice_timer);
2630         cfqg_stats_update_idle_time(cfqq->cfqg);
2631 }
2632
2633 static void __cfq_set_active_queue(struct cfq_data *cfqd,
2634                                    struct cfq_queue *cfqq)
2635 {
2636         if (cfqq) {
2637                 cfq_log_cfqq(cfqd, cfqq, "set_active wl_class:%d wl_type:%d",
2638                                 cfqd->serving_wl_class, cfqd->serving_wl_type);
2639                 cfqg_stats_update_avg_queue_size(cfqq->cfqg);
2640                 cfqq->slice_start = 0;
2641                 cfqq->dispatch_start = ktime_get_ns();
2642                 cfqq->allocated_slice = 0;
2643                 cfqq->slice_end = 0;
2644                 cfqq->slice_dispatch = 0;
2645                 cfqq->nr_sectors = 0;
2646
2647                 cfq_clear_cfqq_wait_request(cfqq);
2648                 cfq_clear_cfqq_must_dispatch(cfqq);
2649                 cfq_clear_cfqq_must_alloc_slice(cfqq);
2650                 cfq_clear_cfqq_fifo_expire(cfqq);
2651                 cfq_mark_cfqq_slice_new(cfqq);
2652
2653                 cfq_del_timer(cfqd, cfqq);
2654         }
2655
2656         cfqd->active_queue = cfqq;
2657 }
2658
2659 /*
2660  * current cfqq expired its slice (or was too idle), select new one
2661  */
2662 static void
2663 __cfq_slice_expired(struct cfq_data *cfqd, struct cfq_queue *cfqq,
2664                     bool timed_out)
2665 {
2666         cfq_log_cfqq(cfqd, cfqq, "slice expired t=%d", timed_out);
2667
2668         if (cfq_cfqq_wait_request(cfqq))
2669                 cfq_del_timer(cfqd, cfqq);
2670
2671         cfq_clear_cfqq_wait_request(cfqq);
2672         cfq_clear_cfqq_wait_busy(cfqq);
2673
2674         /*
2675          * If this cfqq is shared between multiple processes, check to
2676          * make sure that those processes are still issuing I/Os within
2677          * the mean seek distance.  If not, it may be time to break the
2678          * queues apart again.
2679          */
2680         if (cfq_cfqq_coop(cfqq) && CFQQ_SEEKY(cfqq))
2681                 cfq_mark_cfqq_split_coop(cfqq);
2682
2683         /*
2684          * store what was left of this slice, if the queue idled/timed out
2685          */
2686         if (timed_out) {
2687                 if (cfq_cfqq_slice_new(cfqq))
2688                         cfqq->slice_resid = cfq_scaled_cfqq_slice(cfqd, cfqq);
2689                 else
2690                         cfqq->slice_resid = cfqq->slice_end - ktime_get_ns();
2691                 cfq_log_cfqq(cfqd, cfqq, "resid=%lld", cfqq->slice_resid);
2692         }
2693
2694         cfq_group_served(cfqd, cfqq->cfqg, cfqq);
2695
2696         if (cfq_cfqq_on_rr(cfqq) && RB_EMPTY_ROOT(&cfqq->sort_list))
2697                 cfq_del_cfqq_rr(cfqd, cfqq);
2698
2699         cfq_resort_rr_list(cfqd, cfqq);
2700
2701         if (cfqq == cfqd->active_queue)
2702                 cfqd->active_queue = NULL;
2703
2704         if (cfqd->active_cic) {
2705                 put_io_context(cfqd->active_cic->icq.ioc);
2706                 cfqd->active_cic = NULL;
2707         }
2708 }
2709
2710 static inline void cfq_slice_expired(struct cfq_data *cfqd, bool timed_out)
2711 {
2712         struct cfq_queue *cfqq = cfqd->active_queue;
2713
2714         if (cfqq)
2715                 __cfq_slice_expired(cfqd, cfqq, timed_out);
2716 }
2717
2718 /*
2719  * Get next queue for service. Unless we have a queue preemption,
2720  * we'll simply select the first cfqq in the service tree.
2721  */
2722 static struct cfq_queue *cfq_get_next_queue(struct cfq_data *cfqd)
2723 {
2724         struct cfq_rb_root *st = st_for(cfqd->serving_group,
2725                         cfqd->serving_wl_class, cfqd->serving_wl_type);
2726
2727         if (!cfqd->rq_queued)
2728                 return NULL;
2729
2730         /* There is nothing to dispatch */
2731         if (!st)
2732                 return NULL;
2733         if (RB_EMPTY_ROOT(&st->rb))
2734                 return NULL;
2735         return cfq_rb_first(st);
2736 }
2737
2738 static struct cfq_queue *cfq_get_next_queue_forced(struct cfq_data *cfqd)
2739 {
2740         struct cfq_group *cfqg;
2741         struct cfq_queue *cfqq;
2742         int i, j;
2743         struct cfq_rb_root *st;
2744
2745         if (!cfqd->rq_queued)
2746                 return NULL;
2747
2748         cfqg = cfq_get_next_cfqg(cfqd);
2749         if (!cfqg)
2750                 return NULL;
2751
2752         for_each_cfqg_st(cfqg, i, j, st)
2753                 if ((cfqq = cfq_rb_first(st)) != NULL)
2754                         return cfqq;
2755         return NULL;
2756 }
2757
2758 /*
2759  * Get and set a new active queue for service.
2760  */
2761 static struct cfq_queue *cfq_set_active_queue(struct cfq_data *cfqd,
2762                                               struct cfq_queue *cfqq)
2763 {
2764         if (!cfqq)
2765                 cfqq = cfq_get_next_queue(cfqd);
2766
2767         __cfq_set_active_queue(cfqd, cfqq);
2768         return cfqq;
2769 }
2770
2771 static inline sector_t cfq_dist_from_last(struct cfq_data *cfqd,
2772                                           struct request *rq)
2773 {
2774         if (blk_rq_pos(rq) >= cfqd->last_position)
2775                 return blk_rq_pos(rq) - cfqd->last_position;
2776         else
2777                 return cfqd->last_position - blk_rq_pos(rq);
2778 }
2779
2780 static inline int cfq_rq_close(struct cfq_data *cfqd, struct cfq_queue *cfqq,
2781                                struct request *rq)
2782 {
2783         return cfq_dist_from_last(cfqd, rq) <= CFQQ_CLOSE_THR;
2784 }
2785
2786 static struct cfq_queue *cfqq_close(struct cfq_data *cfqd,
2787                                     struct cfq_queue *cur_cfqq)
2788 {
2789         struct rb_root *root = &cfqd->prio_trees[cur_cfqq->org_ioprio];
2790         struct rb_node *parent, *node;
2791         struct cfq_queue *__cfqq;
2792         sector_t sector = cfqd->last_position;
2793
2794         if (RB_EMPTY_ROOT(root))
2795                 return NULL;
2796
2797         /*
2798          * First, if we find a request starting at the end of the last
2799          * request, choose it.
2800          */
2801         __cfqq = cfq_prio_tree_lookup(cfqd, root, sector, &parent, NULL);
2802         if (__cfqq)
2803                 return __cfqq;
2804
2805         /*
2806          * If the exact sector wasn't found, the parent of the NULL leaf
2807          * will contain the closest sector.
2808          */
2809         __cfqq = rb_entry(parent, struct cfq_queue, p_node);
2810         if (cfq_rq_close(cfqd, cur_cfqq, __cfqq->next_rq))
2811                 return __cfqq;
2812
2813         if (blk_rq_pos(__cfqq->next_rq) < sector)
2814                 node = rb_next(&__cfqq->p_node);
2815         else
2816                 node = rb_prev(&__cfqq->p_node);
2817         if (!node)
2818                 return NULL;
2819
2820         __cfqq = rb_entry(node, struct cfq_queue, p_node);
2821         if (cfq_rq_close(cfqd, cur_cfqq, __cfqq->next_rq))
2822                 return __cfqq;
2823
2824         return NULL;
2825 }
2826
2827 /*
2828  * cfqd - obvious
2829  * cur_cfqq - passed in so that we don't decide that the current queue is
2830  *            closely cooperating with itself.
2831  *
2832  * So, basically we're assuming that that cur_cfqq has dispatched at least
2833  * one request, and that cfqd->last_position reflects a position on the disk
2834  * associated with the I/O issued by cur_cfqq.  I'm not sure this is a valid
2835  * assumption.
2836  */
2837 static struct cfq_queue *cfq_close_cooperator(struct cfq_data *cfqd,
2838                                               struct cfq_queue *cur_cfqq)
2839 {
2840         struct cfq_queue *cfqq;
2841
2842         if (cfq_class_idle(cur_cfqq))
2843                 return NULL;
2844         if (!cfq_cfqq_sync(cur_cfqq))
2845                 return NULL;
2846         if (CFQQ_SEEKY(cur_cfqq))
2847                 return NULL;
2848
2849         /*
2850          * Don't search priority tree if it's the only queue in the group.
2851          */
2852         if (cur_cfqq->cfqg->nr_cfqq == 1)
2853                 return NULL;
2854
2855         /*
2856          * We should notice if some of the queues are cooperating, eg
2857          * working closely on the same area of the disk. In that case,
2858          * we can group them together and don't waste time idling.
2859          */
2860         cfqq = cfqq_close(cfqd, cur_cfqq);
2861         if (!cfqq)
2862                 return NULL;
2863
2864         /* If new queue belongs to different cfq_group, don't choose it */
2865         if (cur_cfqq->cfqg != cfqq->cfqg)
2866                 return NULL;
2867
2868         /*
2869          * It only makes sense to merge sync queues.
2870          */
2871         if (!cfq_cfqq_sync(cfqq))
2872                 return NULL;
2873         if (CFQQ_SEEKY(cfqq))
2874                 return NULL;
2875
2876         /*
2877          * Do not merge queues of different priority classes
2878          */
2879         if (cfq_class_rt(cfqq) != cfq_class_rt(cur_cfqq))
2880                 return NULL;
2881
2882         return cfqq;
2883 }
2884
2885 /*
2886  * Determine whether we should enforce idle window for this queue.
2887  */
2888
2889 static bool cfq_should_idle(struct cfq_data *cfqd, struct cfq_queue *cfqq)
2890 {
2891         enum wl_class_t wl_class = cfqq_class(cfqq);
2892         struct cfq_rb_root *st = cfqq->service_tree;
2893
2894         BUG_ON(!st);
2895         BUG_ON(!st->count);
2896
2897         if (!cfqd->cfq_slice_idle)
2898                 return false;
2899
2900         /* We never do for idle class queues. */
2901         if (wl_class == IDLE_WORKLOAD)
2902                 return false;
2903
2904         /* We do for queues that were marked with idle window flag. */
2905         if (cfq_cfqq_idle_window(cfqq) &&
2906            !(blk_queue_nonrot(cfqd->queue) && cfqd->hw_tag))
2907                 return true;
2908
2909         /*
2910          * Otherwise, we do only if they are the last ones
2911          * in their service tree.
2912          */
2913         if (st->count == 1 && cfq_cfqq_sync(cfqq) &&
2914            !cfq_io_thinktime_big(cfqd, &st->ttime, false))
2915                 return true;
2916         cfq_log_cfqq(cfqd, cfqq, "Not idling. st->count:%d", st->count);
2917         return false;
2918 }
2919
2920 static void cfq_arm_slice_timer(struct cfq_data *cfqd)
2921 {
2922         struct cfq_queue *cfqq = cfqd->active_queue;
2923         struct cfq_rb_root *st = cfqq->service_tree;
2924         struct cfq_io_cq *cic;
2925         u64 sl, group_idle = 0;
2926         u64 now = ktime_get_ns();
2927
2928         /*
2929          * SSD device without seek penalty, disable idling. But only do so
2930          * for devices that support queuing, otherwise we still have a problem
2931          * with sync vs async workloads.
2932          */
2933         if (blk_queue_nonrot(cfqd->queue) && cfqd->hw_tag)
2934                 return;
2935
2936         WARN_ON(!RB_EMPTY_ROOT(&cfqq->sort_list));
2937         WARN_ON(cfq_cfqq_slice_new(cfqq));
2938
2939         /*
2940          * idle is disabled, either manually or by past process history
2941          */
2942         if (!cfq_should_idle(cfqd, cfqq)) {
2943                 /* no queue idling. Check for group idling */
2944                 if (cfqd->cfq_group_idle)
2945                         group_idle = cfqd->cfq_group_idle;
2946                 else
2947                         return;
2948         }
2949
2950         /*
2951          * still active requests from this queue, don't idle
2952          */
2953         if (cfqq->dispatched)
2954                 return;
2955
2956         /*
2957          * task has exited, don't wait
2958          */
2959         cic = cfqd->active_cic;
2960         if (!cic || !atomic_read(&cic->icq.ioc->active_ref))
2961                 return;
2962
2963         /*
2964          * If our average think time is larger than the remaining time
2965          * slice, then don't idle. This avoids overrunning the allotted
2966          * time slice.
2967          */
2968         if (sample_valid(cic->ttime.ttime_samples) &&
2969             (cfqq->slice_end - now < cic->ttime.ttime_mean)) {
2970                 cfq_log_cfqq(cfqd, cfqq, "Not idling. think_time:%llu",
2971                              cic->ttime.ttime_mean);
2972                 return;
2973         }
2974
2975         /*
2976          * There are other queues in the group or this is the only group and
2977          * it has too big thinktime, don't do group idle.
2978          */
2979         if (group_idle &&
2980             (cfqq->cfqg->nr_cfqq > 1 ||
2981              cfq_io_thinktime_big(cfqd, &st->ttime, true)))
2982                 return;
2983
2984         cfq_mark_cfqq_wait_request(cfqq);
2985
2986         if (group_idle)
2987                 sl = cfqd->cfq_group_idle;
2988         else
2989                 sl = cfqd->cfq_slice_idle;
2990
2991         hrtimer_start(&cfqd->idle_slice_timer, ns_to_ktime(sl),
2992                       HRTIMER_MODE_REL);
2993         cfqg_stats_set_start_idle_time(cfqq->cfqg);
2994         cfq_log_cfqq(cfqd, cfqq, "arm_idle: %llu group_idle: %d", sl,
2995                         group_idle ? 1 : 0);
2996 }
2997
2998 /*
2999  * Move request from internal lists to the request queue dispatch list.
3000  */
3001 static void cfq_dispatch_insert(struct request_queue *q, struct request *rq)
3002 {
3003         struct cfq_data *cfqd = q->elevator->elevator_data;
3004         struct cfq_queue *cfqq = RQ_CFQQ(rq);
3005
3006         cfq_log_cfqq(cfqd, cfqq, "dispatch_insert");
3007
3008         cfqq->next_rq = cfq_find_next_rq(cfqd, cfqq, rq);
3009         cfq_remove_request(rq);
3010         cfqq->dispatched++;
3011         (RQ_CFQG(rq))->dispatched++;
3012         elv_dispatch_sort(q, rq);
3013
3014         cfqd->rq_in_flight[cfq_cfqq_sync(cfqq)]++;
3015         cfqq->nr_sectors += blk_rq_sectors(rq);
3016 }
3017
3018 /*
3019  * return expired entry, or NULL to just start from scratch in rbtree
3020  */
3021 static struct request *cfq_check_fifo(struct cfq_queue *cfqq)
3022 {
3023         struct request *rq = NULL;
3024
3025         if (cfq_cfqq_fifo_expire(cfqq))
3026                 return NULL;
3027
3028         cfq_mark_cfqq_fifo_expire(cfqq);
3029
3030         if (list_empty(&cfqq->fifo))
3031                 return NULL;
3032
3033         rq = rq_entry_fifo(cfqq->fifo.next);
3034         if (ktime_get_ns() < rq->fifo_time)
3035                 rq = NULL;
3036
3037         return rq;
3038 }
3039
3040 static inline int
3041 cfq_prio_to_maxrq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
3042 {
3043         const int base_rq = cfqd->cfq_slice_async_rq;
3044
3045         WARN_ON(cfqq->ioprio >= IOPRIO_BE_NR);
3046
3047         return 2 * base_rq * (IOPRIO_BE_NR - cfqq->ioprio);
3048 }
3049
3050 /*
3051  * Must be called with the queue_lock held.
3052  */
3053 static int cfqq_process_refs(struct cfq_queue *cfqq)
3054 {
3055         int process_refs, io_refs;
3056
3057         io_refs = cfqq->allocated[READ] + cfqq->allocated[WRITE];
3058         process_refs = cfqq->ref - io_refs;
3059         BUG_ON(process_refs < 0);
3060         return process_refs;
3061 }
3062
3063 static void cfq_setup_merge(struct cfq_queue *cfqq, struct cfq_queue *new_cfqq)
3064 {
3065         int process_refs, new_process_refs;
3066         struct cfq_queue *__cfqq;
3067
3068         /*
3069          * If there are no process references on the new_cfqq, then it is
3070          * unsafe to follow the ->new_cfqq chain as other cfqq's in the
3071          * chain may have dropped their last reference (not just their
3072          * last process reference).
3073          */
3074         if (!cfqq_process_refs(new_cfqq))
3075                 return;
3076
3077         /* Avoid a circular list and skip interim queue merges */
3078         while ((__cfqq = new_cfqq->new_cfqq)) {
3079                 if (__cfqq == cfqq)
3080                         return;
3081                 new_cfqq = __cfqq;
3082         }
3083
3084         process_refs = cfqq_process_refs(cfqq);
3085         new_process_refs = cfqq_process_refs(new_cfqq);
3086         /*
3087          * If the process for the cfqq has gone away, there is no
3088          * sense in merging the queues.
3089          */
3090         if (process_refs == 0 || new_process_refs == 0)
3091                 return;
3092
3093         /*
3094          * Merge in the direction of the lesser amount of work.
3095          */
3096         if (new_process_refs >= process_refs) {
3097                 cfqq->new_cfqq = new_cfqq;
3098                 new_cfqq->ref += process_refs;
3099         } else {
3100                 new_cfqq->new_cfqq = cfqq;
3101                 cfqq->ref += new_process_refs;
3102         }
3103 }
3104
3105 static enum wl_type_t cfq_choose_wl_type(struct cfq_data *cfqd,
3106                         struct cfq_group *cfqg, enum wl_class_t wl_class)
3107 {
3108         struct cfq_queue *queue;
3109         int i;
3110         bool key_valid = false;
3111         u64 lowest_key = 0;
3112         enum wl_type_t cur_best = SYNC_NOIDLE_WORKLOAD;
3113
3114         for (i = 0; i <= SYNC_WORKLOAD; ++i) {
3115                 /* select the one with lowest rb_key */
3116                 queue = cfq_rb_first(st_for(cfqg, wl_class, i));
3117                 if (queue &&
3118                     (!key_valid || queue->rb_key < lowest_key)) {
3119                         lowest_key = queue->rb_key;
3120                         cur_best = i;
3121                         key_valid = true;
3122                 }
3123         }
3124
3125         return cur_best;
3126 }
3127
3128 static void
3129 choose_wl_class_and_type(struct cfq_data *cfqd, struct cfq_group *cfqg)
3130 {
3131         u64 slice;
3132         unsigned count;
3133         struct cfq_rb_root *st;
3134         u64 group_slice;
3135         enum wl_class_t original_class = cfqd->serving_wl_class;
3136         u64 now = ktime_get_ns();
3137
3138         /* Choose next priority. RT > BE > IDLE */
3139         if (cfq_group_busy_queues_wl(RT_WORKLOAD, cfqd, cfqg))
3140                 cfqd->serving_wl_class = RT_WORKLOAD;
3141         else if (cfq_group_busy_queues_wl(BE_WORKLOAD, cfqd, cfqg))
3142                 cfqd->serving_wl_class = BE_WORKLOAD;
3143         else {
3144                 cfqd->serving_wl_class = IDLE_WORKLOAD;
3145                 cfqd->workload_expires = now + jiffies_to_nsecs(1);
3146                 return;
3147         }
3148
3149         if (original_class != cfqd->serving_wl_class)
3150                 goto new_workload;
3151
3152         /*
3153          * For RT and BE, we have to choose also the type
3154          * (SYNC, SYNC_NOIDLE, ASYNC), and to compute a workload
3155          * expiration time
3156          */
3157         st = st_for(cfqg, cfqd->serving_wl_class, cfqd->serving_wl_type);
3158         count = st->count;
3159
3160         /*
3161          * check workload expiration, and that we still have other queues ready
3162          */
3163         if (count && !(now > cfqd->workload_expires))
3164                 return;
3165
3166 new_workload:
3167         /* otherwise select new workload type */
3168         cfqd->serving_wl_type = cfq_choose_wl_type(cfqd, cfqg,
3169                                         cfqd->serving_wl_class);
3170         st = st_for(cfqg, cfqd->serving_wl_class, cfqd->serving_wl_type);
3171         count = st->count;
3172
3173         /*
3174          * the workload slice is computed as a fraction of target latency
3175          * proportional to the number of queues in that workload, over
3176          * all the queues in the same priority class
3177          */
3178         group_slice = cfq_group_slice(cfqd, cfqg);
3179
3180         slice = div_u64(group_slice * count,
3181                 max_t(unsigned, cfqg->busy_queues_avg[cfqd->serving_wl_class],
3182                       cfq_group_busy_queues_wl(cfqd->serving_wl_class, cfqd,
3183                                         cfqg)));
3184
3185         if (cfqd->serving_wl_type == ASYNC_WORKLOAD) {
3186                 u64 tmp;
3187
3188                 /*
3189                  * Async queues are currently system wide. Just taking
3190                  * proportion of queues with-in same group will lead to higher
3191                  * async ratio system wide as generally root group is going
3192                  * to have higher weight. A more accurate thing would be to
3193                  * calculate system wide asnc/sync ratio.
3194                  */
3195                 tmp = cfqd->cfq_target_latency *
3196                         cfqg_busy_async_queues(cfqd, cfqg);
3197                 tmp = div_u64(tmp, cfqd->busy_queues);
3198                 slice = min_t(u64, slice, tmp);
3199
3200                 /* async workload slice is scaled down according to
3201                  * the sync/async slice ratio. */
3202                 slice = div64_u64(slice*cfqd->cfq_slice[0], cfqd->cfq_slice[1]);
3203         } else
3204                 /* sync workload slice is at least 2 * cfq_slice_idle */
3205                 slice = max(slice, 2 * cfqd->cfq_slice_idle);
3206
3207         slice = max_t(u64, slice, CFQ_MIN_TT);
3208         cfq_log(cfqd, "workload slice:%llu", slice);
3209         cfqd->workload_expires = now + slice;
3210 }
3211
3212 static struct cfq_group *cfq_get_next_cfqg(struct cfq_data *cfqd)
3213 {
3214         struct cfq_rb_root *st = &cfqd->grp_service_tree;
3215         struct cfq_group *cfqg;
3216
3217         if (RB_EMPTY_ROOT(&st->rb))
3218                 return NULL;
3219         cfqg = cfq_rb_first_group(st);
3220         update_min_vdisktime(st);
3221         return cfqg;
3222 }
3223
3224 static void cfq_choose_cfqg(struct cfq_data *cfqd)
3225 {
3226         struct cfq_group *cfqg = cfq_get_next_cfqg(cfqd);
3227         u64 now = ktime_get_ns();
3228
3229         cfqd->serving_group = cfqg;
3230
3231         /* Restore the workload type data */
3232         if (cfqg->saved_wl_slice) {
3233                 cfqd->workload_expires = now + cfqg->saved_wl_slice;
3234                 cfqd->serving_wl_type = cfqg->saved_wl_type;
3235                 cfqd->serving_wl_class = cfqg->saved_wl_class;
3236         } else
3237                 cfqd->workload_expires = now - 1;
3238
3239         choose_wl_class_and_type(cfqd, cfqg);
3240 }
3241
3242 /*
3243  * Select a queue for service. If we have a current active queue,
3244  * check whether to continue servicing it, or retrieve and set a new one.
3245  */
3246 static struct cfq_queue *cfq_select_queue(struct cfq_data *cfqd)
3247 {
3248         struct cfq_queue *cfqq, *new_cfqq = NULL;
3249         u64 now = ktime_get_ns();
3250
3251         cfqq = cfqd->active_queue;
3252         if (!cfqq)
3253                 goto new_queue;
3254
3255         if (!cfqd->rq_queued)
3256                 return NULL;
3257
3258         /*
3259          * We were waiting for group to get backlogged. Expire the queue
3260          */
3261         if (cfq_cfqq_wait_busy(cfqq) && !RB_EMPTY_ROOT(&cfqq->sort_list))
3262                 goto expire;
3263
3264         /*
3265          * The active queue has run out of time, expire it and select new.
3266          */
3267         if (cfq_slice_used(cfqq) && !cfq_cfqq_must_dispatch(cfqq)) {
3268                 /*
3269                  * If slice had not expired at the completion of last request
3270                  * we might not have turned on wait_busy flag. Don't expire
3271                  * the queue yet. Allow the group to get backlogged.
3272                  *
3273                  * The very fact that we have used the slice, that means we
3274                  * have been idling all along on this queue and it should be
3275                  * ok to wait for this request to complete.
3276                  */
3277                 if (cfqq->cfqg->nr_cfqq == 1 && RB_EMPTY_ROOT(&cfqq->sort_list)
3278                     && cfqq->dispatched && cfq_should_idle(cfqd, cfqq)) {
3279                         cfqq = NULL;
3280                         goto keep_queue;
3281                 } else
3282                         goto check_group_idle;
3283         }
3284
3285         /*
3286          * The active queue has requests and isn't expired, allow it to
3287          * dispatch.
3288          */
3289         if (!RB_EMPTY_ROOT(&cfqq->sort_list))
3290                 goto keep_queue;
3291
3292         /*
3293          * If another queue has a request waiting within our mean seek
3294          * distance, let it run.  The expire code will check for close
3295          * cooperators and put the close queue at the front of the service
3296          * tree.  If possible, merge the expiring queue with the new cfqq.
3297          */
3298         new_cfqq = cfq_close_cooperator(cfqd, cfqq);
3299         if (new_cfqq) {
3300                 if (!cfqq->new_cfqq)
3301                         cfq_setup_merge(cfqq, new_cfqq);
3302                 goto expire;
3303         }
3304
3305         /*
3306          * No requests pending. If the active queue still has requests in
3307          * flight or is idling for a new request, allow either of these
3308          * conditions to happen (or time out) before selecting a new queue.
3309          */
3310         if (hrtimer_active(&cfqd->idle_slice_timer)) {
3311                 cfqq = NULL;
3312                 goto keep_queue;
3313         }
3314
3315         /*
3316          * This is a deep seek queue, but the device is much faster than
3317          * the queue can deliver, don't idle
3318          **/
3319         if (CFQQ_SEEKY(cfqq) && cfq_cfqq_idle_window(cfqq) &&
3320             (cfq_cfqq_slice_new(cfqq) ||
3321             (cfqq->slice_end - now > now - cfqq->slice_start))) {
3322                 cfq_clear_cfqq_deep(cfqq);
3323                 cfq_clear_cfqq_idle_window(cfqq);
3324         }
3325
3326         if (cfqq->dispatched && cfq_should_idle(cfqd, cfqq)) {
3327                 cfqq = NULL;
3328                 goto keep_queue;
3329         }
3330
3331         /*
3332          * If group idle is enabled and there are requests dispatched from
3333          * this group, wait for requests to complete.
3334          */
3335 check_group_idle:
3336         if (cfqd->cfq_group_idle && cfqq->cfqg->nr_cfqq == 1 &&
3337             cfqq->cfqg->dispatched &&
3338             !cfq_io_thinktime_big(cfqd, &cfqq->cfqg->ttime, true)) {
3339                 cfqq = NULL;
3340                 goto keep_queue;
3341         }
3342
3343 expire:
3344         cfq_slice_expired(cfqd, 0);
3345 new_queue:
3346         /*
3347          * Current queue expired. Check if we have to switch to a new
3348          * service tree
3349          */
3350         if (!new_cfqq)
3351                 cfq_choose_cfqg(cfqd);
3352
3353         cfqq = cfq_set_active_queue(cfqd, new_cfqq);
3354 keep_queue:
3355         return cfqq;
3356 }
3357
3358 static int __cfq_forced_dispatch_cfqq(struct cfq_queue *cfqq)
3359 {
3360         int dispatched = 0;
3361
3362         while (cfqq->next_rq) {
3363                 cfq_dispatch_insert(cfqq->cfqd->queue, cfqq->next_rq);
3364                 dispatched++;
3365         }
3366
3367         BUG_ON(!list_empty(&cfqq->fifo));
3368
3369         /* By default cfqq is not expired if it is empty. Do it explicitly */
3370         __cfq_slice_expired(cfqq->cfqd, cfqq, 0);
3371         return dispatched;
3372 }
3373
3374 /*
3375  * Drain our current requests. Used for barriers and when switching
3376  * io schedulers on-the-fly.
3377  */
3378 static int cfq_forced_dispatch(struct cfq_data *cfqd)
3379 {
3380         struct cfq_queue *cfqq;
3381         int dispatched = 0;
3382
3383         /* Expire the timeslice of the current active queue first */
3384         cfq_slice_expired(cfqd, 0);
3385         while ((cfqq = cfq_get_next_queue_forced(cfqd)) != NULL) {
3386                 __cfq_set_active_queue(cfqd, cfqq);
3387                 dispatched += __cfq_forced_dispatch_cfqq(cfqq);
3388         }
3389
3390         BUG_ON(cfqd->busy_queues);
3391
3392         cfq_log(cfqd, "forced_dispatch=%d", dispatched);
3393         return dispatched;
3394 }
3395
3396 static inline bool cfq_slice_used_soon(struct cfq_data *cfqd,
3397         struct cfq_queue *cfqq)
3398 {
3399         u64 now = ktime_get_ns();
3400
3401         /* the queue hasn't finished any request, can't estimate */
3402         if (cfq_cfqq_slice_new(cfqq))
3403                 return true;
3404         if (now + cfqd->cfq_slice_idle * cfqq->dispatched > cfqq->slice_end)
3405                 return true;
3406
3407         return false;
3408 }
3409
3410 static bool cfq_may_dispatch(struct cfq_data *cfqd, struct cfq_queue *cfqq)
3411 {
3412         unsigned int max_dispatch;
3413
3414         if (cfq_cfqq_must_dispatch(cfqq))
3415                 return true;
3416
3417         /*
3418          * Drain async requests before we start sync IO
3419          */
3420         if (cfq_should_idle(cfqd, cfqq) && cfqd->rq_in_flight[BLK_RW_ASYNC])
3421                 return false;
3422
3423         /*
3424          * If this is an async queue and we have sync IO in flight, let it wait
3425          */
3426         if (cfqd->rq_in_flight[BLK_RW_SYNC] && !cfq_cfqq_sync(cfqq))
3427                 return false;
3428
3429         max_dispatch = max_t(unsigned int, cfqd->cfq_quantum / 2, 1);
3430         if (cfq_class_idle(cfqq))
3431                 max_dispatch = 1;
3432
3433         /*
3434          * Does this cfqq already have too much IO in flight?
3435          */
3436         if (cfqq->dispatched >= max_dispatch) {
3437                 bool promote_sync = false;
3438                 /*
3439                  * idle queue must always only have a single IO in flight
3440                  */
3441                 if (cfq_class_idle(cfqq))
3442                         return false;
3443
3444                 /*
3445                  * If there is only one sync queue
3446                  * we can ignore async queue here and give the sync
3447                  * queue no dispatch limit. The reason is a sync queue can
3448                  * preempt async queue, limiting the sync queue doesn't make
3449                  * sense. This is useful for aiostress test.
3450                  */
3451                 if (cfq_cfqq_sync(cfqq) && cfqd->busy_sync_queues == 1)
3452                         promote_sync = true;
3453
3454                 /*
3455                  * We have other queues, don't allow more IO from this one
3456                  */
3457                 if (cfqd->busy_queues > 1 && cfq_slice_used_soon(cfqd, cfqq) &&
3458                                 !promote_sync)
3459                         return false;
3460
3461                 /*
3462                  * Sole queue user, no limit
3463                  */
3464                 if (cfqd->busy_queues == 1 || promote_sync)
3465                         max_dispatch = -1;
3466                 else
3467                         /*
3468                          * Normally we start throttling cfqq when cfq_quantum/2
3469                          * requests have been dispatched. But we can drive
3470                          * deeper queue depths at the beginning of slice
3471                          * subjected to upper limit of cfq_quantum.
3472                          * */
3473                         max_dispatch = cfqd->cfq_quantum;
3474         }
3475
3476         /*
3477          * Async queues must wait a bit before being allowed dispatch.
3478          * We also ramp up the dispatch depth gradually for async IO,
3479          * based on the last sync IO we serviced
3480          */
3481         if (!cfq_cfqq_sync(cfqq) && cfqd->cfq_latency) {
3482                 u64 last_sync = ktime_get_ns() - cfqd->last_delayed_sync;
3483                 unsigned int depth;
3484
3485                 depth = div64_u64(last_sync, cfqd->cfq_slice[1]);
3486                 if (!depth && !cfqq->dispatched)
3487                         depth = 1;
3488                 if (depth < max_dispatch)
3489                         max_dispatch = depth;
3490         }
3491
3492         /*
3493          * If we're below the current max, allow a dispatch
3494          */
3495         return cfqq->dispatched < max_dispatch;
3496 }
3497
3498 /*
3499  * Dispatch a request from cfqq, moving them to the request queue
3500  * dispatch list.
3501  */
3502 static bool cfq_dispatch_request(struct cfq_data *cfqd, struct cfq_queue *cfqq)
3503 {
3504         struct request *rq;
3505
3506         BUG_ON(RB_EMPTY_ROOT(&cfqq->sort_list));
3507
3508         rq = cfq_check_fifo(cfqq);
3509         if (rq)
3510                 cfq_mark_cfqq_must_dispatch(cfqq);
3511
3512         if (!cfq_may_dispatch(cfqd, cfqq))
3513                 return false;
3514
3515         /*
3516          * follow expired path, else get first next available
3517          */
3518         if (!rq)
3519                 rq = cfqq->next_rq;
3520         else
3521                 cfq_log_cfqq(cfqq->cfqd, cfqq, "fifo=%p", rq);
3522
3523         /*
3524          * insert request into driver dispatch list
3525          */
3526         cfq_dispatch_insert(cfqd->queue, rq);
3527
3528         if (!cfqd->active_cic) {
3529                 struct cfq_io_cq *cic = RQ_CIC(rq);
3530
3531                 atomic_long_inc(&cic->icq.ioc->refcount);
3532                 cfqd->active_cic = cic;
3533         }
3534
3535         return true;
3536 }
3537
3538 /*
3539  * Find the cfqq that we need to service and move a request from that to the
3540  * dispatch list
3541  */
3542 static int cfq_dispatch_requests(struct request_queue *q, int force)
3543 {
3544         struct cfq_data *cfqd = q->elevator->elevator_data;
3545         struct cfq_queue *cfqq;
3546
3547         if (!cfqd->busy_queues)
3548                 return 0;
3549
3550         if (unlikely(force))
3551                 return cfq_forced_dispatch(cfqd);
3552
3553         cfqq = cfq_select_queue(cfqd);
3554         if (!cfqq)
3555                 return 0;
3556
3557         /*
3558          * Dispatch a request from this cfqq, if it is allowed
3559          */
3560         if (!cfq_dispatch_request(cfqd, cfqq))
3561                 return 0;
3562
3563         cfqq->slice_dispatch++;
3564         cfq_clear_cfqq_must_dispatch(cfqq);
3565
3566         /*
3567          * expire an async queue immediately if it has used up its slice. idle
3568          * queue always expire after 1 dispatch round.
3569          */
3570         if (cfqd->busy_queues > 1 && ((!cfq_cfqq_sync(cfqq) &&
3571             cfqq->slice_dispatch >= cfq_prio_to_maxrq(cfqd, cfqq)) ||
3572             cfq_class_idle(cfqq))) {
3573                 cfqq->slice_end = ktime_get_ns() + 1;
3574                 cfq_slice_expired(cfqd, 0);
3575         }
3576
3577         cfq_log_cfqq(cfqd, cfqq, "dispatched a request");
3578         return 1;
3579 }
3580
3581 /*
3582  * task holds one reference to the queue, dropped when task exits. each rq
3583  * in-flight on this queue also holds a reference, dropped when rq is freed.
3584  *
3585  * Each cfq queue took a reference on the parent group. Drop it now.
3586  * queue lock must be held here.
3587  */
3588 static void cfq_put_queue(struct cfq_queue *cfqq)
3589 {
3590         struct cfq_data *cfqd = cfqq->cfqd;
3591         struct cfq_group *cfqg;
3592
3593         BUG_ON(cfqq->ref <= 0);
3594
3595         cfqq->ref--;
3596         if (cfqq->ref)
3597                 return;
3598
3599         cfq_log_cfqq(cfqd, cfqq, "put_queue");
3600         BUG_ON(rb_first(&cfqq->sort_list));
3601         BUG_ON(cfqq->allocated[READ] + cfqq->allocated[WRITE]);
3602         cfqg = cfqq->cfqg;
3603
3604         if (unlikely(cfqd->active_queue == cfqq)) {
3605                 __cfq_slice_expired(cfqd, cfqq, 0);
3606                 cfq_schedule_dispatch(cfqd);
3607         }
3608
3609         BUG_ON(cfq_cfqq_on_rr(cfqq));
3610         kmem_cache_free(cfq_pool, cfqq);
3611         cfqg_put(cfqg);
3612 }
3613
3614 static void cfq_put_cooperator(struct cfq_queue *cfqq)
3615 {
3616         struct cfq_queue *__cfqq, *next;
3617
3618         /*
3619          * If this queue was scheduled to merge with another queue, be
3620          * sure to drop the reference taken on that queue (and others in
3621          * the merge chain).  See cfq_setup_merge and cfq_merge_cfqqs.
3622          */
3623         __cfqq = cfqq->new_cfqq;
3624         while (__cfqq) {
3625                 if (__cfqq == cfqq) {
3626                         WARN(1, "cfqq->new_cfqq loop detected\n");
3627                         break;
3628                 }
3629                 next = __cfqq->new_cfqq;
3630                 cfq_put_queue(__cfqq);
3631                 __cfqq = next;
3632         }
3633 }
3634
3635 static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
3636 {
3637         if (unlikely(cfqq == cfqd->active_queue)) {
3638                 __cfq_slice_expired(cfqd, cfqq, 0);
3639                 cfq_schedule_dispatch(cfqd);
3640         }
3641
3642         cfq_put_cooperator(cfqq);
3643
3644         cfq_put_queue(cfqq);
3645 }
3646
3647 static void cfq_init_icq(struct io_cq *icq)
3648 {
3649         struct cfq_io_cq *cic = icq_to_cic(icq);
3650
3651         cic->ttime.last_end_request = ktime_get_ns();
3652 }
3653
3654 static void cfq_exit_icq(struct io_cq *icq)
3655 {
3656         struct cfq_io_cq *cic = icq_to_cic(icq);
3657         struct cfq_data *cfqd = cic_to_cfqd(cic);
3658
3659         if (cic_to_cfqq(cic, false)) {
3660                 cfq_exit_cfqq(cfqd, cic_to_cfqq(cic, false));
3661                 cic_set_cfqq(cic, NULL, false);
3662         }
3663
3664         if (cic_to_cfqq(cic, true)) {
3665                 cfq_exit_cfqq(cfqd, cic_to_cfqq(cic, true));
3666                 cic_set_cfqq(cic, NULL, true);
3667         }
3668 }
3669
3670 static void cfq_init_prio_data(struct cfq_queue *cfqq, struct cfq_io_cq *cic)
3671 {
3672         struct task_struct *tsk = current;
3673         int ioprio_class;
3674
3675         if (!cfq_cfqq_prio_changed(cfqq))
3676                 return;
3677
3678         ioprio_class = IOPRIO_PRIO_CLASS(cic->ioprio);
3679         switch (ioprio_class) {
3680         default:
3681                 printk(KERN_ERR "cfq: bad prio %x\n", ioprio_class);
3682         case IOPRIO_CLASS_NONE:
3683                 /*
3684                  * no prio set, inherit CPU scheduling settings
3685                  */
3686                 cfqq->ioprio = task_nice_ioprio(tsk);
3687                 cfqq->ioprio_class = task_nice_ioclass(tsk);
3688                 break;
3689         case IOPRIO_CLASS_RT:
3690                 cfqq->ioprio = IOPRIO_PRIO_DATA(cic->ioprio);
3691                 cfqq->ioprio_class = IOPRIO_CLASS_RT;
3692                 break;
3693         case IOPRIO_CLASS_BE:
3694                 cfqq->ioprio = IOPRIO_PRIO_DATA(cic->ioprio);
3695                 cfqq->ioprio_class = IOPRIO_CLASS_BE;
3696                 break;
3697         case IOPRIO_CLASS_IDLE:
3698                 cfqq->ioprio_class = IOPRIO_CLASS_IDLE;
3699                 cfqq->ioprio = 7;
3700                 cfq_clear_cfqq_idle_window(cfqq);
3701                 break;
3702         }
3703
3704         /*
3705          * keep track of original prio settings in case we have to temporarily
3706          * elevate the priority of this queue
3707          */
3708         cfqq->org_ioprio = cfqq->ioprio;
3709         cfqq->org_ioprio_class = cfqq->ioprio_class;
3710         cfq_clear_cfqq_prio_changed(cfqq);
3711 }
3712
3713 static void check_ioprio_changed(struct cfq_io_cq *cic, struct bio *bio)
3714 {
3715         int ioprio = cic->icq.ioc->ioprio;
3716         struct cfq_data *cfqd = cic_to_cfqd(cic);
3717         struct cfq_queue *cfqq;
3718
3719         /*
3720          * Check whether ioprio has changed.  The condition may trigger
3721          * spuriously on a newly created cic but there's no harm.
3722          */
3723         if (unlikely(!cfqd) || likely(cic->ioprio == ioprio))
3724                 return;
3725
3726         cfqq = cic_to_cfqq(cic, false);
3727         if (cfqq) {
3728                 cfq_put_queue(cfqq);
3729                 cfqq = cfq_get_queue(cfqd, BLK_RW_ASYNC, cic, bio);
3730                 cic_set_cfqq(cic, cfqq, false);
3731         }
3732
3733         cfqq = cic_to_cfqq(cic, true);
3734         if (cfqq)
3735                 cfq_mark_cfqq_prio_changed(cfqq);
3736
3737         cic->ioprio = ioprio;
3738 }
3739
3740 static void cfq_init_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq,
3741                           pid_t pid, bool is_sync)
3742 {
3743         RB_CLEAR_NODE(&cfqq->rb_node);
3744         RB_CLEAR_NODE(&cfqq->p_node);
3745         INIT_LIST_HEAD(&cfqq->fifo);
3746
3747         cfqq->ref = 0;
3748         cfqq->cfqd = cfqd;
3749
3750         cfq_mark_cfqq_prio_changed(cfqq);
3751
3752         if (is_sync) {
3753                 if (!cfq_class_idle(cfqq))
3754                         cfq_mark_cfqq_idle_window(cfqq);
3755                 cfq_mark_cfqq_sync(cfqq);
3756         }
3757         cfqq->pid = pid;
3758 }
3759
3760 #ifdef CONFIG_CFQ_GROUP_IOSCHED
3761 static void check_blkcg_changed(struct cfq_io_cq *cic, struct bio *bio)
3762 {
3763         struct cfq_data *cfqd = cic_to_cfqd(cic);
3764         struct cfq_queue *cfqq;
3765         uint64_t serial_nr;
3766         bool nonroot_cg;
3767
3768         rcu_read_lock();
3769         serial_nr = bio_blkcg(bio)->css.serial_nr;
3770         nonroot_cg = bio_blkcg(bio) != &blkcg_root;
3771         rcu_read_unlock();
3772
3773         /*
3774          * Check whether blkcg has changed.  The condition may trigger
3775          * spuriously on a newly created cic but there's no harm.
3776          */
3777         if (unlikely(!cfqd) || likely(cic->blkcg_serial_nr == serial_nr))
3778                 return;
3779
3780         /*
3781          * If we have a non-root cgroup, we can depend on that to
3782          * do proper throttling of writes. Turn off wbt for that
3783          * case, if it was enabled by default.
3784          */
3785         if (nonroot_cg)
3786                 wbt_disable_default(cfqd->queue);
3787
3788         /*
3789          * Drop reference to queues.  New queues will be assigned in new
3790          * group upon arrival of fresh requests.
3791          */
3792         cfqq = cic_to_cfqq(cic, false);
3793         if (cfqq) {
3794                 cfq_log_cfqq(cfqd, cfqq, "changed cgroup");
3795                 cic_set_cfqq(cic, NULL, false);
3796                 cfq_put_queue(cfqq);
3797         }
3798
3799         cfqq = cic_to_cfqq(cic, true);
3800         if (cfqq) {
3801                 cfq_log_cfqq(cfqd, cfqq, "changed cgroup");
3802                 cic_set_cfqq(cic, NULL, true);
3803                 cfq_put_queue(cfqq);
3804         }
3805
3806         cic->blkcg_serial_nr = serial_nr;
3807 }
3808 #else
3809 static inline void check_blkcg_changed(struct cfq_io_cq *cic, struct bio *bio) { }
3810 #endif  /* CONFIG_CFQ_GROUP_IOSCHED */
3811
3812 static struct cfq_queue **
3813 cfq_async_queue_prio(struct cfq_group *cfqg, int ioprio_class, int ioprio)
3814 {
3815         switch (ioprio_class) {
3816         case IOPRIO_CLASS_RT:
3817                 return &cfqg->async_cfqq[0][ioprio];
3818         case IOPRIO_CLASS_NONE:
3819                 ioprio = IOPRIO_NORM;
3820                 /* fall through */
3821         case IOPRIO_CLASS_BE:
3822                 return &cfqg->async_cfqq[1][ioprio];
3823         case IOPRIO_CLASS_IDLE:
3824                 return &cfqg->async_idle_cfqq;
3825         default:
3826                 BUG();
3827         }
3828 }
3829
3830 static struct cfq_queue *
3831 cfq_get_queue(struct cfq_data *cfqd, bool is_sync, struct cfq_io_cq *cic,
3832               struct bio *bio)
3833 {
3834         int ioprio_class = IOPRIO_PRIO_CLASS(cic->ioprio);
3835         int ioprio = IOPRIO_PRIO_DATA(cic->ioprio);
3836         struct cfq_queue **async_cfqq = NULL;
3837         struct cfq_queue *cfqq;
3838         struct cfq_group *cfqg;
3839
3840         rcu_read_lock();
3841         cfqg = cfq_lookup_cfqg(cfqd, bio_blkcg(bio));
3842         if (!cfqg) {
3843                 cfqq = &cfqd->oom_cfqq;
3844                 goto out;
3845         }
3846
3847         if (!is_sync) {
3848                 if (!ioprio_valid(cic->ioprio)) {
3849                         struct task_struct *tsk = current;
3850                         ioprio = task_nice_ioprio(tsk);
3851                         ioprio_class = task_nice_ioclass(tsk);
3852                 }
3853                 async_cfqq = cfq_async_queue_prio(cfqg, ioprio_class, ioprio);
3854                 cfqq = *async_cfqq;
3855                 if (cfqq)
3856                         goto out;
3857         }
3858
3859         cfqq = kmem_cache_alloc_node(cfq_pool,
3860                                      GFP_NOWAIT | __GFP_ZERO | __GFP_NOWARN,
3861                                      cfqd->queue->node);
3862         if (!cfqq) {
3863                 cfqq = &cfqd->oom_cfqq;
3864                 goto out;
3865         }
3866
3867         cfq_init_cfqq(cfqd, cfqq, current->pid, is_sync);
3868         cfq_init_prio_data(cfqq, cic);
3869         cfq_link_cfqq_cfqg(cfqq, cfqg);
3870         cfq_log_cfqq(cfqd, cfqq, "alloced");
3871
3872         if (async_cfqq) {
3873                 /* a new async queue is created, pin and remember */
3874                 cfqq->ref++;
3875                 *async_cfqq = cfqq;
3876         }
3877 out:
3878         cfqq->ref++;
3879         rcu_read_unlock();
3880         return cfqq;
3881 }
3882
3883 static void
3884 __cfq_update_io_thinktime(struct cfq_ttime *ttime, u64 slice_idle)
3885 {
3886         u64 elapsed = ktime_get_ns() - ttime->last_end_request;
3887         elapsed = min(elapsed, 2UL * slice_idle);
3888
3889         ttime->ttime_samples = (7*ttime->ttime_samples + 256) / 8;
3890         ttime->ttime_total = div_u64(7*ttime->ttime_total + 256*elapsed,  8);
3891         ttime->ttime_mean = div64_ul(ttime->ttime_total + 128,
3892                                      ttime->ttime_samples);
3893 }
3894
3895 static void
3896 cfq_update_io_thinktime(struct cfq_data *cfqd, struct cfq_queue *cfqq,
3897                         struct cfq_io_cq *cic)
3898 {
3899         if (cfq_cfqq_sync(cfqq)) {
3900                 __cfq_update_io_thinktime(&cic->ttime, cfqd->cfq_slice_idle);
3901                 __cfq_update_io_thinktime(&cfqq->service_tree->ttime,
3902                         cfqd->cfq_slice_idle);
3903         }
3904 #ifdef CONFIG_CFQ_GROUP_IOSCHED
3905         __cfq_update_io_thinktime(&cfqq->cfqg->ttime, cfqd->cfq_group_idle);
3906 #endif
3907 }
3908
3909 static void
3910 cfq_update_io_seektime(struct cfq_data *cfqd, struct cfq_queue *cfqq,
3911                        struct request *rq)
3912 {
3913         sector_t sdist = 0;
3914         sector_t n_sec = blk_rq_sectors(rq);
3915         if (cfqq->last_request_pos) {
3916                 if (cfqq->last_request_pos < blk_rq_pos(rq))
3917                         sdist = blk_rq_pos(rq) - cfqq->last_request_pos;
3918                 else
3919                         sdist = cfqq->last_request_pos - blk_rq_pos(rq);
3920         }
3921
3922         cfqq->seek_history <<= 1;
3923         if (blk_queue_nonrot(cfqd->queue))
3924                 cfqq->seek_history |= (n_sec < CFQQ_SECT_THR_NONROT);
3925         else
3926                 cfqq->seek_history |= (sdist > CFQQ_SEEK_THR);
3927 }
3928
3929 static inline bool req_noidle(struct request *req)
3930 {
3931         return req_op(req) == REQ_OP_WRITE &&
3932                 (req->cmd_flags & (REQ_SYNC | REQ_IDLE)) == REQ_SYNC;
3933 }
3934
3935 /*
3936  * Disable idle window if the process thinks too long or seeks so much that
3937  * it doesn't matter
3938  */
3939 static void
3940 cfq_update_idle_window(struct cfq_data *cfqd, struct cfq_queue *cfqq,
3941                        struct cfq_io_cq *cic)
3942 {
3943         int old_idle, enable_idle;
3944
3945         /*
3946          * Don't idle for async or idle io prio class
3947          */
3948         if (!cfq_cfqq_sync(cfqq) || cfq_class_idle(cfqq))
3949                 return;
3950
3951         enable_idle = old_idle = cfq_cfqq_idle_window(cfqq);
3952
3953         if (cfqq->queued[0] + cfqq->queued[1] >= 4)
3954                 cfq_mark_cfqq_deep(cfqq);
3955
3956         if (cfqq->next_rq && req_noidle(cfqq->next_rq))
3957                 enable_idle = 0;
3958         else if (!atomic_read(&cic->icq.ioc->active_ref) ||
3959                  !cfqd->cfq_slice_idle ||
3960                  (!cfq_cfqq_deep(cfqq) && CFQQ_SEEKY(cfqq)))
3961                 enable_idle = 0;
3962         else if (sample_valid(cic->ttime.ttime_samples)) {
3963                 if (cic->ttime.ttime_mean > cfqd->cfq_slice_idle)
3964                         enable_idle = 0;
3965                 else
3966                         enable_idle = 1;
3967         }
3968
3969         if (old_idle != enable_idle) {
3970                 cfq_log_cfqq(cfqd, cfqq, "idle=%d", enable_idle);
3971                 if (enable_idle)
3972                         cfq_mark_cfqq_idle_window(cfqq);
3973                 else
3974                         cfq_clear_cfqq_idle_window(cfqq);
3975         }
3976 }
3977
3978 /*
3979  * Check if new_cfqq should preempt the currently active queue. Return 0 for
3980  * no or if we aren't sure, a 1 will cause a preempt.
3981  */
3982 static bool
3983 cfq_should_preempt(struct cfq_data *cfqd, struct cfq_queue *new_cfqq,
3984                    struct request *rq)
3985 {
3986         struct cfq_queue *cfqq;
3987
3988         cfqq = cfqd->active_queue;
3989         if (!cfqq)
3990                 return false;
3991
3992         if (cfq_class_idle(new_cfqq))
3993                 return false;
3994
3995         if (cfq_class_idle(cfqq))
3996                 return true;
3997
3998         /*
3999          * Don't allow a non-RT request to preempt an ongoing RT cfqq timeslice.
4000          */
4001         if (cfq_class_rt(cfqq) && !cfq_class_rt(new_cfqq))
4002                 return false;
4003
4004         /*
4005          * if the new request is sync, but the currently running queue is
4006          * not, let the sync request have priority.
4007          */
4008         if (rq_is_sync(rq) && !cfq_cfqq_sync(cfqq) && !cfq_cfqq_must_dispatch(cfqq))
4009                 return true;
4010
4011         /*
4012          * Treat ancestors of current cgroup the same way as current cgroup.
4013          * For anybody else we disallow preemption to guarantee service
4014          * fairness among cgroups.
4015          */
4016         if (!cfqg_is_descendant(cfqq->cfqg, new_cfqq->cfqg))
4017                 return false;
4018
4019         if (cfq_slice_used(cfqq))
4020                 return true;
4021
4022         /*
4023          * Allow an RT request to pre-empt an ongoing non-RT cfqq timeslice.
4024          */
4025         if (cfq_class_rt(new_cfqq) && !cfq_class_rt(cfqq))
4026                 return true;
4027
4028         WARN_ON_ONCE(cfqq->ioprio_class != new_cfqq->ioprio_class);
4029         /* Allow preemption only if we are idling on sync-noidle tree */
4030         if (cfqd->serving_wl_type == SYNC_NOIDLE_WORKLOAD &&
4031             cfqq_type(new_cfqq) == SYNC_NOIDLE_WORKLOAD &&
4032             RB_EMPTY_ROOT(&cfqq->sort_list))
4033                 return true;
4034
4035         /*
4036          * So both queues are sync. Let the new request get disk time if
4037          * it's a metadata request and the current queue is doing regular IO.
4038          */
4039         if ((rq->cmd_flags & REQ_PRIO) && !cfqq->prio_pending)
4040                 return true;
4041
4042         /* An idle queue should not be idle now for some reason */
4043         if (RB_EMPTY_ROOT(&cfqq->sort_list) && !cfq_should_idle(cfqd, cfqq))
4044                 return true;
4045
4046         if (!cfqd->active_cic || !cfq_cfqq_wait_request(cfqq))
4047                 return false;
4048
4049         /*
4050          * if this request is as-good as one we would expect from the
4051          * current cfqq, let it preempt
4052          */
4053         if (cfq_rq_close(cfqd, cfqq, rq))
4054                 return true;
4055
4056         return false;
4057 }
4058
4059 /*
4060  * cfqq preempts the active queue. if we allowed preempt with no slice left,
4061  * let it have half of its nominal slice.
4062  */
4063 static void cfq_preempt_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq)
4064 {
4065         enum wl_type_t old_type = cfqq_type(cfqd->active_queue);
4066
4067         cfq_log_cfqq(cfqd, cfqq, "preempt");
4068         cfq_slice_expired(cfqd, 1);
4069
4070         /*
4071          * workload type is changed, don't save slice, otherwise preempt
4072          * doesn't happen
4073          */
4074         if (old_type != cfqq_type(cfqq))
4075                 cfqq->cfqg->saved_wl_slice = 0;
4076
4077         /*
4078          * Put the new queue at the front of the of the current list,
4079          * so we know that it will be selected next.
4080          */
4081         BUG_ON(!cfq_cfqq_on_rr(cfqq));
4082
4083         cfq_service_tree_add(cfqd, cfqq, 1);
4084
4085         cfqq->slice_end = 0;
4086         cfq_mark_cfqq_slice_new(cfqq);
4087 }
4088
4089 /*
4090  * Called when a new fs request (rq) is added (to cfqq). Check if there's
4091  * something we should do about it
4092  */
4093 static void
4094 cfq_rq_enqueued(struct cfq_data *cfqd, struct cfq_queue *cfqq,
4095                 struct request *rq)
4096 {
4097         struct cfq_io_cq *cic = RQ_CIC(rq);
4098
4099         cfqd->rq_queued++;
4100         if (rq->cmd_flags & REQ_PRIO)
4101                 cfqq->prio_pending++;
4102
4103         cfq_update_io_thinktime(cfqd, cfqq, cic);
4104         cfq_update_io_seektime(cfqd, cfqq, rq);
4105         cfq_update_idle_window(cfqd, cfqq, cic);
4106
4107         cfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq);
4108
4109         if (cfqq == cfqd->active_queue) {
4110                 /*
4111                  * Remember that we saw a request from this process, but
4112                  * don't start queuing just yet. Otherwise we risk seeing lots
4113                  * of tiny requests, because we disrupt the normal plugging
4114                  * and merging. If the request is already larger than a single
4115                  * page, let it rip immediately. For that case we assume that
4116                  * merging is already done. Ditto for a busy system that
4117                  * has other work pending, don't risk delaying until the
4118                  * idle timer unplug to continue working.
4119                  */
4120                 if (cfq_cfqq_wait_request(cfqq)) {
4121                         if (blk_rq_bytes(rq) > PAGE_SIZE ||
4122                             cfqd->busy_queues > 1) {
4123                                 cfq_del_timer(cfqd, cfqq);
4124                                 cfq_clear_cfqq_wait_request(cfqq);
4125                                 __blk_run_queue(cfqd->queue);
4126                         } else {
4127                                 cfqg_stats_update_idle_time(cfqq->cfqg);
4128                                 cfq_mark_cfqq_must_dispatch(cfqq);
4129                         }
4130                 }
4131         } else if (cfq_should_preempt(cfqd, cfqq, rq)) {
4132                 /*
4133                  * not the active queue - expire current slice if it is
4134                  * idle and has expired it's mean thinktime or this new queue
4135                  * has some old slice time left and is of higher priority or
4136                  * this new queue is RT and the current one is BE
4137                  */
4138                 cfq_preempt_queue(cfqd, cfqq);
4139                 __blk_run_queue(cfqd->queue);
4140         }
4141 }
4142
4143 static void cfq_insert_request(struct request_queue *q, struct request *rq)
4144 {
4145         struct cfq_data *cfqd = q->elevator->elevator_data;
4146         struct cfq_queue *cfqq = RQ_CFQQ(rq);
4147
4148         cfq_log_cfqq(cfqd, cfqq, "insert_request");
4149         cfq_init_prio_data(cfqq, RQ_CIC(rq));
4150
4151         rq->fifo_time = ktime_get_ns() + cfqd->cfq_fifo_expire[rq_is_sync(rq)];
4152         list_add_tail(&rq->queuelist, &cfqq->fifo);
4153         cfq_add_rq_rb(rq);
4154         cfqg_stats_update_io_add(RQ_CFQG(rq), cfqd->serving_group,
4155                                  rq->cmd_flags);
4156         cfq_rq_enqueued(cfqd, cfqq, rq);
4157 }
4158
4159 /*
4160  * Update hw_tag based on peak queue depth over 50 samples under
4161  * sufficient load.
4162  */
4163 static void cfq_update_hw_tag(struct cfq_data *cfqd)
4164 {
4165         struct cfq_queue *cfqq = cfqd->active_queue;
4166
4167         if (cfqd->rq_in_driver > cfqd->hw_tag_est_depth)
4168                 cfqd->hw_tag_est_depth = cfqd->rq_in_driver;
4169
4170         if (cfqd->hw_tag == 1)
4171                 return;
4172
4173         if (cfqd->rq_queued <= CFQ_HW_QUEUE_MIN &&
4174             cfqd->rq_in_driver <= CFQ_HW_QUEUE_MIN)
4175                 return;
4176
4177         /*
4178          * If active queue hasn't enough requests and can idle, cfq might not
4179          * dispatch sufficient requests to hardware. Don't zero hw_tag in this
4180          * case
4181          */
4182         if (cfqq && cfq_cfqq_idle_window(cfqq) &&
4183             cfqq->dispatched + cfqq->queued[0] + cfqq->queued[1] <
4184             CFQ_HW_QUEUE_MIN && cfqd->rq_in_driver < CFQ_HW_QUEUE_MIN)
4185                 return;
4186
4187         if (cfqd->hw_tag_samples++ < 50)
4188                 return;
4189
4190         if (cfqd->hw_tag_est_depth >= CFQ_HW_QUEUE_MIN)
4191                 cfqd->hw_tag = 1;
4192         else
4193                 cfqd->hw_tag = 0;
4194 }
4195
4196 static bool cfq_should_wait_busy(struct cfq_data *cfqd, struct cfq_queue *cfqq)
4197 {
4198         struct cfq_io_cq *cic = cfqd->active_cic;
4199         u64 now = ktime_get_ns();
4200
4201         /* If the queue already has requests, don't wait */
4202         if (!RB_EMPTY_ROOT(&cfqq->sort_list))
4203                 return false;
4204
4205         /* If there are other queues in the group, don't wait */
4206         if (cfqq->cfqg->nr_cfqq > 1)
4207                 return false;
4208
4209         /* the only queue in the group, but think time is big */
4210         if (cfq_io_thinktime_big(cfqd, &cfqq->cfqg->ttime, true))
4211                 return false;
4212
4213         if (cfq_slice_used(cfqq))
4214                 return true;
4215
4216         /* if slice left is less than think time, wait busy */
4217         if (cic && sample_valid(cic->ttime.ttime_samples)
4218             && (cfqq->slice_end - now < cic->ttime.ttime_mean))
4219                 return true;
4220
4221         /*
4222          * If think times is less than a jiffy than ttime_mean=0 and above
4223          * will not be true. It might happen that slice has not expired yet
4224          * but will expire soon (4-5 ns) during select_queue(). To cover the
4225          * case where think time is less than a jiffy, mark the queue wait
4226          * busy if only 1 jiffy is left in the slice.
4227          */
4228         if (cfqq->slice_end - now <= jiffies_to_nsecs(1))
4229                 return true;
4230
4231         return false;
4232 }
4233
4234 static void cfq_completed_request(struct request_queue *q, struct request *rq)
4235 {
4236         struct cfq_queue *cfqq = RQ_CFQQ(rq);
4237         struct cfq_data *cfqd = cfqq->cfqd;
4238         const int sync = rq_is_sync(rq);
4239         u64 now = ktime_get_ns();
4240
4241         cfq_log_cfqq(cfqd, cfqq, "complete rqnoidle %d", req_noidle(rq));
4242
4243         cfq_update_hw_tag(cfqd);
4244
4245         WARN_ON(!cfqd->rq_in_driver);
4246         WARN_ON(!cfqq->dispatched);
4247         cfqd->rq_in_driver--;
4248         cfqq->dispatched--;
4249         (RQ_CFQG(rq))->dispatched--;
4250         cfqg_stats_update_completion(cfqq->cfqg, rq_start_time_ns(rq),
4251                                      rq_io_start_time_ns(rq), rq->cmd_flags);
4252
4253         cfqd->rq_in_flight[cfq_cfqq_sync(cfqq)]--;
4254
4255         if (sync) {
4256                 struct cfq_rb_root *st;
4257
4258                 RQ_CIC(rq)->ttime.last_end_request = now;
4259
4260                 if (cfq_cfqq_on_rr(cfqq))
4261                         st = cfqq->service_tree;
4262                 else
4263                         st = st_for(cfqq->cfqg, cfqq_class(cfqq),
4264                                         cfqq_type(cfqq));
4265
4266                 st->ttime.last_end_request = now;
4267                 /*
4268                  * We have to do this check in jiffies since start_time is in
4269                  * jiffies and it is not trivial to convert to ns. If
4270                  * cfq_fifo_expire[1] ever comes close to 1 jiffie, this test
4271                  * will become problematic but so far we are fine (the default
4272                  * is 128 ms).
4273                  */
4274                 if (!time_after(rq->start_time +
4275                                   nsecs_to_jiffies(cfqd->cfq_fifo_expire[1]),
4276                                 jiffies))
4277                         cfqd->last_delayed_sync = now;
4278         }
4279
4280 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4281         cfqq->cfqg->ttime.last_end_request = now;
4282 #endif
4283
4284         /*
4285          * If this is the active queue, check if it needs to be expired,
4286          * or if we want to idle in case it has no pending requests.
4287          */
4288         if (cfqd->active_queue == cfqq) {
4289                 const bool cfqq_empty = RB_EMPTY_ROOT(&cfqq->sort_list);
4290
4291                 if (cfq_cfqq_slice_new(cfqq)) {
4292                         cfq_set_prio_slice(cfqd, cfqq);
4293                         cfq_clear_cfqq_slice_new(cfqq);
4294                 }
4295
4296                 /*
4297                  * Should we wait for next request to come in before we expire
4298                  * the queue.
4299                  */
4300                 if (cfq_should_wait_busy(cfqd, cfqq)) {
4301                         u64 extend_sl = cfqd->cfq_slice_idle;
4302                         if (!cfqd->cfq_slice_idle)
4303                                 extend_sl = cfqd->cfq_group_idle;
4304                         cfqq->slice_end = now + extend_sl;
4305                         cfq_mark_cfqq_wait_busy(cfqq);
4306                         cfq_log_cfqq(cfqd, cfqq, "will busy wait");
4307                 }
4308
4309                 /*
4310                  * Idling is not enabled on:
4311                  * - expired queues
4312                  * - idle-priority queues
4313                  * - async queues
4314                  * - queues with still some requests queued
4315                  * - when there is a close cooperator
4316                  */
4317                 if (cfq_slice_used(cfqq) || cfq_class_idle(cfqq))
4318                         cfq_slice_expired(cfqd, 1);
4319                 else if (sync && cfqq_empty &&
4320                          !cfq_close_cooperator(cfqd, cfqq)) {
4321                         cfq_arm_slice_timer(cfqd);
4322                 }
4323         }
4324
4325         if (!cfqd->rq_in_driver)
4326                 cfq_schedule_dispatch(cfqd);
4327 }
4328
4329 static void cfqq_boost_on_prio(struct cfq_queue *cfqq, unsigned int op)
4330 {
4331         /*
4332          * If REQ_PRIO is set, boost class and prio level, if it's below
4333          * BE/NORM. If prio is not set, restore the potentially boosted
4334          * class/prio level.
4335          */
4336         if (!(op & REQ_PRIO)) {
4337                 cfqq->ioprio_class = cfqq->org_ioprio_class;
4338                 cfqq->ioprio = cfqq->org_ioprio;
4339         } else {
4340                 if (cfq_class_idle(cfqq))
4341                         cfqq->ioprio_class = IOPRIO_CLASS_BE;
4342                 if (cfqq->ioprio > IOPRIO_NORM)
4343                         cfqq->ioprio = IOPRIO_NORM;
4344         }
4345 }
4346
4347 static inline int __cfq_may_queue(struct cfq_queue *cfqq)
4348 {
4349         if (cfq_cfqq_wait_request(cfqq) && !cfq_cfqq_must_alloc_slice(cfqq)) {
4350                 cfq_mark_cfqq_must_alloc_slice(cfqq);
4351                 return ELV_MQUEUE_MUST;
4352         }
4353
4354         return ELV_MQUEUE_MAY;
4355 }
4356
4357 static int cfq_may_queue(struct request_queue *q, unsigned int op)
4358 {
4359         struct cfq_data *cfqd = q->elevator->elevator_data;
4360         struct task_struct *tsk = current;
4361         struct cfq_io_cq *cic;
4362         struct cfq_queue *cfqq;
4363
4364         /*
4365          * don't force setup of a queue from here, as a call to may_queue
4366          * does not necessarily imply that a request actually will be queued.
4367          * so just lookup a possibly existing queue, or return 'may queue'
4368          * if that fails
4369          */
4370         cic = cfq_cic_lookup(cfqd, tsk->io_context);
4371         if (!cic)
4372                 return ELV_MQUEUE_MAY;
4373
4374         cfqq = cic_to_cfqq(cic, op_is_sync(op));
4375         if (cfqq) {
4376                 cfq_init_prio_data(cfqq, cic);
4377                 cfqq_boost_on_prio(cfqq, op);
4378
4379                 return __cfq_may_queue(cfqq);
4380         }
4381
4382         return ELV_MQUEUE_MAY;
4383 }
4384
4385 /*
4386  * queue lock held here
4387  */
4388 static void cfq_put_request(struct request *rq)
4389 {
4390         struct cfq_queue *cfqq = RQ_CFQQ(rq);
4391
4392         if (cfqq) {
4393                 const int rw = rq_data_dir(rq);
4394
4395                 BUG_ON(!cfqq->allocated[rw]);
4396                 cfqq->allocated[rw]--;
4397
4398                 /* Put down rq reference on cfqg */
4399                 cfqg_put(RQ_CFQG(rq));
4400                 rq->elv.priv[0] = NULL;
4401                 rq->elv.priv[1] = NULL;
4402
4403                 cfq_put_queue(cfqq);
4404         }
4405 }
4406
4407 static struct cfq_queue *
4408 cfq_merge_cfqqs(struct cfq_data *cfqd, struct cfq_io_cq *cic,
4409                 struct cfq_queue *cfqq)
4410 {
4411         cfq_log_cfqq(cfqd, cfqq, "merging with queue %p", cfqq->new_cfqq);
4412         cic_set_cfqq(cic, cfqq->new_cfqq, 1);
4413         cfq_mark_cfqq_coop(cfqq->new_cfqq);
4414         cfq_put_queue(cfqq);
4415         return cic_to_cfqq(cic, 1);
4416 }
4417
4418 /*
4419  * Returns NULL if a new cfqq should be allocated, or the old cfqq if this
4420  * was the last process referring to said cfqq.
4421  */
4422 static struct cfq_queue *
4423 split_cfqq(struct cfq_io_cq *cic, struct cfq_queue *cfqq)
4424 {
4425         if (cfqq_process_refs(cfqq) == 1) {
4426                 cfqq->pid = current->pid;
4427                 cfq_clear_cfqq_coop(cfqq);
4428                 cfq_clear_cfqq_split_coop(cfqq);
4429                 return cfqq;
4430         }
4431
4432         cic_set_cfqq(cic, NULL, 1);
4433
4434         cfq_put_cooperator(cfqq);
4435
4436         cfq_put_queue(cfqq);
4437         return NULL;
4438 }
4439 /*
4440  * Allocate cfq data structures associated with this request.
4441  */
4442 static int
4443 cfq_set_request(struct request_queue *q, struct request *rq, struct bio *bio,
4444                 gfp_t gfp_mask)
4445 {
4446         struct cfq_data *cfqd = q->elevator->elevator_data;
4447         struct cfq_io_cq *cic = icq_to_cic(rq->elv.icq);
4448         const int rw = rq_data_dir(rq);
4449         const bool is_sync = rq_is_sync(rq);
4450         struct cfq_queue *cfqq;
4451
4452         spin_lock_irq(q->queue_lock);
4453
4454         check_ioprio_changed(cic, bio);
4455         check_blkcg_changed(cic, bio);
4456 new_queue:
4457         cfqq = cic_to_cfqq(cic, is_sync);
4458         if (!cfqq || cfqq == &cfqd->oom_cfqq) {
4459                 if (cfqq)
4460                         cfq_put_queue(cfqq);
4461                 cfqq = cfq_get_queue(cfqd, is_sync, cic, bio);
4462                 cic_set_cfqq(cic, cfqq, is_sync);
4463         } else {
4464                 /*
4465                  * If the queue was seeky for too long, break it apart.
4466                  */
4467                 if (cfq_cfqq_coop(cfqq) && cfq_cfqq_split_coop(cfqq)) {
4468                         cfq_log_cfqq(cfqd, cfqq, "breaking apart cfqq");
4469                         cfqq = split_cfqq(cic, cfqq);
4470                         if (!cfqq)
4471                                 goto new_queue;
4472                 }
4473
4474                 /*
4475                  * Check to see if this queue is scheduled to merge with
4476                  * another, closely cooperating queue.  The merging of
4477                  * queues happens here as it must be done in process context.
4478                  * The reference on new_cfqq was taken in merge_cfqqs.
4479                  */
4480                 if (cfqq->new_cfqq)
4481                         cfqq = cfq_merge_cfqqs(cfqd, cic, cfqq);
4482         }
4483
4484         cfqq->allocated[rw]++;
4485
4486         cfqq->ref++;
4487         cfqg_get(cfqq->cfqg);
4488         rq->elv.priv[0] = cfqq;
4489         rq->elv.priv[1] = cfqq->cfqg;
4490         spin_unlock_irq(q->queue_lock);
4491         return 0;
4492 }
4493
4494 static void cfq_kick_queue(struct work_struct *work)
4495 {
4496         struct cfq_data *cfqd =
4497                 container_of(work, struct cfq_data, unplug_work);
4498         struct request_queue *q = cfqd->queue;
4499
4500         spin_lock_irq(q->queue_lock);
4501         __blk_run_queue(cfqd->queue);
4502         spin_unlock_irq(q->queue_lock);
4503 }
4504
4505 /*
4506  * Timer running if the active_queue is currently idling inside its time slice
4507  */
4508 static enum hrtimer_restart cfq_idle_slice_timer(struct hrtimer *timer)
4509 {
4510         struct cfq_data *cfqd = container_of(timer, struct cfq_data,
4511                                              idle_slice_timer);
4512         struct cfq_queue *cfqq;
4513         unsigned long flags;
4514         int timed_out = 1;
4515
4516         cfq_log(cfqd, "idle timer fired");
4517
4518         spin_lock_irqsave(cfqd->queue->queue_lock, flags);
4519
4520         cfqq = cfqd->active_queue;
4521         if (cfqq) {
4522                 timed_out = 0;
4523
4524                 /*
4525                  * We saw a request before the queue expired, let it through
4526                  */
4527                 if (cfq_cfqq_must_dispatch(cfqq))
4528                         goto out_kick;
4529
4530                 /*
4531                  * expired
4532                  */
4533                 if (cfq_slice_used(cfqq))
4534                         goto expire;
4535
4536                 /*
4537                  * only expire and reinvoke request handler, if there are
4538                  * other queues with pending requests
4539                  */
4540                 if (!cfqd->busy_queues)
4541                         goto out_cont;
4542
4543                 /*
4544                  * not expired and it has a request pending, let it dispatch
4545                  */
4546                 if (!RB_EMPTY_ROOT(&cfqq->sort_list))
4547                         goto out_kick;
4548
4549                 /*
4550                  * Queue depth flag is reset only when the idle didn't succeed
4551                  */
4552                 cfq_clear_cfqq_deep(cfqq);
4553         }
4554 expire:
4555         cfq_slice_expired(cfqd, timed_out);
4556 out_kick:
4557         cfq_schedule_dispatch(cfqd);
4558 out_cont:
4559         spin_unlock_irqrestore(cfqd->queue->queue_lock, flags);
4560         return HRTIMER_NORESTART;
4561 }
4562
4563 static void cfq_shutdown_timer_wq(struct cfq_data *cfqd)
4564 {
4565         hrtimer_cancel(&cfqd->idle_slice_timer);
4566         cancel_work_sync(&cfqd->unplug_work);
4567 }
4568
4569 static void cfq_exit_queue(struct elevator_queue *e)
4570 {
4571         struct cfq_data *cfqd = e->elevator_data;
4572         struct request_queue *q = cfqd->queue;
4573
4574         cfq_shutdown_timer_wq(cfqd);
4575
4576         spin_lock_irq(q->queue_lock);
4577
4578         if (cfqd->active_queue)
4579                 __cfq_slice_expired(cfqd, cfqd->active_queue, 0);
4580
4581         spin_unlock_irq(q->queue_lock);
4582
4583         cfq_shutdown_timer_wq(cfqd);
4584
4585 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4586         blkcg_deactivate_policy(q, &blkcg_policy_cfq);
4587 #else
4588         kfree(cfqd->root_group);
4589 #endif
4590         kfree(cfqd);
4591 }
4592
4593 static int cfq_init_queue(struct request_queue *q, struct elevator_type *e)
4594 {
4595         struct cfq_data *cfqd;
4596         struct blkcg_gq *blkg __maybe_unused;
4597         int i, ret;
4598         struct elevator_queue *eq;
4599
4600         eq = elevator_alloc(q, e);
4601         if (!eq)
4602                 return -ENOMEM;
4603
4604         cfqd = kzalloc_node(sizeof(*cfqd), GFP_KERNEL, q->node);
4605         if (!cfqd) {
4606                 kobject_put(&eq->kobj);
4607                 return -ENOMEM;
4608         }
4609         eq->elevator_data = cfqd;
4610
4611         cfqd->queue = q;
4612         spin_lock_irq(q->queue_lock);
4613         q->elevator = eq;
4614         spin_unlock_irq(q->queue_lock);
4615
4616         /* Init root service tree */
4617         cfqd->grp_service_tree = CFQ_RB_ROOT;
4618
4619         /* Init root group and prefer root group over other groups by default */
4620 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4621         ret = blkcg_activate_policy(q, &blkcg_policy_cfq);
4622         if (ret)
4623                 goto out_free;
4624
4625         cfqd->root_group = blkg_to_cfqg(q->root_blkg);
4626 #else
4627         ret = -ENOMEM;
4628         cfqd->root_group = kzalloc_node(sizeof(*cfqd->root_group),
4629                                         GFP_KERNEL, cfqd->queue->node);
4630         if (!cfqd->root_group)
4631                 goto out_free;
4632
4633         cfq_init_cfqg_base(cfqd->root_group);
4634         cfqd->root_group->weight = 2 * CFQ_WEIGHT_LEGACY_DFL;
4635         cfqd->root_group->leaf_weight = 2 * CFQ_WEIGHT_LEGACY_DFL;
4636 #endif
4637
4638         /*
4639          * Not strictly needed (since RB_ROOT just clears the node and we
4640          * zeroed cfqd on alloc), but better be safe in case someone decides
4641          * to add magic to the rb code
4642          */
4643         for (i = 0; i < CFQ_PRIO_LISTS; i++)
4644                 cfqd->prio_trees[i] = RB_ROOT;
4645
4646         /*
4647          * Our fallback cfqq if cfq_get_queue() runs into OOM issues.
4648          * Grab a permanent reference to it, so that the normal code flow
4649          * will not attempt to free it.  oom_cfqq is linked to root_group
4650          * but shouldn't hold a reference as it'll never be unlinked.  Lose
4651          * the reference from linking right away.
4652          */
4653         cfq_init_cfqq(cfqd, &cfqd->oom_cfqq, 1, 0);
4654         cfqd->oom_cfqq.ref++;
4655
4656         spin_lock_irq(q->queue_lock);
4657         cfq_link_cfqq_cfqg(&cfqd->oom_cfqq, cfqd->root_group);
4658         cfqg_put(cfqd->root_group);
4659         spin_unlock_irq(q->queue_lock);
4660
4661         hrtimer_init(&cfqd->idle_slice_timer, CLOCK_MONOTONIC,
4662                      HRTIMER_MODE_REL);
4663         cfqd->idle_slice_timer.function = cfq_idle_slice_timer;
4664
4665         INIT_WORK(&cfqd->unplug_work, cfq_kick_queue);
4666
4667         cfqd->cfq_quantum = cfq_quantum;
4668         cfqd->cfq_fifo_expire[0] = cfq_fifo_expire[0];
4669         cfqd->cfq_fifo_expire[1] = cfq_fifo_expire[1];
4670         cfqd->cfq_back_max = cfq_back_max;
4671         cfqd->cfq_back_penalty = cfq_back_penalty;
4672         cfqd->cfq_slice[0] = cfq_slice_async;
4673         cfqd->cfq_slice[1] = cfq_slice_sync;
4674         cfqd->cfq_target_latency = cfq_target_latency;
4675         cfqd->cfq_slice_async_rq = cfq_slice_async_rq;
4676         cfqd->cfq_slice_idle = cfq_slice_idle;
4677         cfqd->cfq_group_idle = cfq_group_idle;
4678         cfqd->cfq_latency = 1;
4679         cfqd->hw_tag = -1;
4680         /*
4681          * we optimistically start assuming sync ops weren't delayed in last
4682          * second, in order to have larger depth for async operations.
4683          */
4684         cfqd->last_delayed_sync = ktime_get_ns() - NSEC_PER_SEC;
4685         return 0;
4686
4687 out_free:
4688         kfree(cfqd);
4689         kobject_put(&eq->kobj);
4690         return ret;
4691 }
4692
4693 static void cfq_registered_queue(struct request_queue *q)
4694 {
4695         struct elevator_queue *e = q->elevator;
4696         struct cfq_data *cfqd = e->elevator_data;
4697
4698         /*
4699          * Default to IOPS mode with no idling for SSDs
4700          */
4701         if (blk_queue_nonrot(q))
4702                 cfqd->cfq_slice_idle = 0;
4703 }
4704
4705 /*
4706  * sysfs parts below -->
4707  */
4708 static ssize_t
4709 cfq_var_show(unsigned int var, char *page)
4710 {
4711         return sprintf(page, "%u\n", var);
4712 }
4713
4714 static ssize_t
4715 cfq_var_store(unsigned int *var, const char *page, size_t count)
4716 {
4717         char *p = (char *) page;
4718
4719         *var = simple_strtoul(p, &p, 10);
4720         return count;
4721 }
4722
4723 #define SHOW_FUNCTION(__FUNC, __VAR, __CONV)                            \
4724 static ssize_t __FUNC(struct elevator_queue *e, char *page)             \
4725 {                                                                       \
4726         struct cfq_data *cfqd = e->elevator_data;                       \
4727         u64 __data = __VAR;                                             \
4728         if (__CONV)                                                     \
4729                 __data = div_u64(__data, NSEC_PER_MSEC);                        \
4730         return cfq_var_show(__data, (page));                            \
4731 }
4732 SHOW_FUNCTION(cfq_quantum_show, cfqd->cfq_quantum, 0);
4733 SHOW_FUNCTION(cfq_fifo_expire_sync_show, cfqd->cfq_fifo_expire[1], 1);
4734 SHOW_FUNCTION(cfq_fifo_expire_async_show, cfqd->cfq_fifo_expire[0], 1);
4735 SHOW_FUNCTION(cfq_back_seek_max_show, cfqd->cfq_back_max, 0);
4736 SHOW_FUNCTION(cfq_back_seek_penalty_show, cfqd->cfq_back_penalty, 0);
4737 SHOW_FUNCTION(cfq_slice_idle_show, cfqd->cfq_slice_idle, 1);
4738 SHOW_FUNCTION(cfq_group_idle_show, cfqd->cfq_group_idle, 1);
4739 SHOW_FUNCTION(cfq_slice_sync_show, cfqd->cfq_slice[1], 1);
4740 SHOW_FUNCTION(cfq_slice_async_show, cfqd->cfq_slice[0], 1);
4741 SHOW_FUNCTION(cfq_slice_async_rq_show, cfqd->cfq_slice_async_rq, 0);
4742 SHOW_FUNCTION(cfq_low_latency_show, cfqd->cfq_latency, 0);
4743 SHOW_FUNCTION(cfq_target_latency_show, cfqd->cfq_target_latency, 1);
4744 #undef SHOW_FUNCTION
4745
4746 #define USEC_SHOW_FUNCTION(__FUNC, __VAR)                               \
4747 static ssize_t __FUNC(struct elevator_queue *e, char *page)             \
4748 {                                                                       \
4749         struct cfq_data *cfqd = e->elevator_data;                       \
4750         u64 __data = __VAR;                                             \
4751         __data = div_u64(__data, NSEC_PER_USEC);                        \
4752         return cfq_var_show(__data, (page));                            \
4753 }
4754 USEC_SHOW_FUNCTION(cfq_slice_idle_us_show, cfqd->cfq_slice_idle);
4755 USEC_SHOW_FUNCTION(cfq_group_idle_us_show, cfqd->cfq_group_idle);
4756 USEC_SHOW_FUNCTION(cfq_slice_sync_us_show, cfqd->cfq_slice[1]);
4757 USEC_SHOW_FUNCTION(cfq_slice_async_us_show, cfqd->cfq_slice[0]);
4758 USEC_SHOW_FUNCTION(cfq_target_latency_us_show, cfqd->cfq_target_latency);
4759 #undef USEC_SHOW_FUNCTION
4760
4761 #define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX, __CONV)                 \
4762 static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count) \
4763 {                                                                       \
4764         struct cfq_data *cfqd = e->elevator_data;                       \
4765         unsigned int __data;                                            \
4766         int ret = cfq_var_store(&__data, (page), count);                \
4767         if (__data < (MIN))                                             \
4768                 __data = (MIN);                                         \
4769         else if (__data > (MAX))                                        \
4770                 __data = (MAX);                                         \
4771         if (__CONV)                                                     \
4772                 *(__PTR) = (u64)__data * NSEC_PER_MSEC;                 \
4773         else                                                            \
4774                 *(__PTR) = __data;                                      \
4775         return ret;                                                     \
4776 }
4777 STORE_FUNCTION(cfq_quantum_store, &cfqd->cfq_quantum, 1, UINT_MAX, 0);
4778 STORE_FUNCTION(cfq_fifo_expire_sync_store, &cfqd->cfq_fifo_expire[1], 1,
4779                 UINT_MAX, 1);
4780 STORE_FUNCTION(cfq_fifo_expire_async_store, &cfqd->cfq_fifo_expire[0], 1,
4781                 UINT_MAX, 1);
4782 STORE_FUNCTION(cfq_back_seek_max_store, &cfqd->cfq_back_max, 0, UINT_MAX, 0);
4783 STORE_FUNCTION(cfq_back_seek_penalty_store, &cfqd->cfq_back_penalty, 1,
4784                 UINT_MAX, 0);
4785 STORE_FUNCTION(cfq_slice_idle_store, &cfqd->cfq_slice_idle, 0, UINT_MAX, 1);
4786 STORE_FUNCTION(cfq_group_idle_store, &cfqd->cfq_group_idle, 0, UINT_MAX, 1);
4787 STORE_FUNCTION(cfq_slice_sync_store, &cfqd->cfq_slice[1], 1, UINT_MAX, 1);
4788 STORE_FUNCTION(cfq_slice_async_store, &cfqd->cfq_slice[0], 1, UINT_MAX, 1);
4789 STORE_FUNCTION(cfq_slice_async_rq_store, &cfqd->cfq_slice_async_rq, 1,
4790                 UINT_MAX, 0);
4791 STORE_FUNCTION(cfq_low_latency_store, &cfqd->cfq_latency, 0, 1, 0);
4792 STORE_FUNCTION(cfq_target_latency_store, &cfqd->cfq_target_latency, 1, UINT_MAX, 1);
4793 #undef STORE_FUNCTION
4794
4795 #define USEC_STORE_FUNCTION(__FUNC, __PTR, MIN, MAX)                    \
4796 static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count) \
4797 {                                                                       \
4798         struct cfq_data *cfqd = e->elevator_data;                       \
4799         unsigned int __data;                                            \
4800         int ret = cfq_var_store(&__data, (page), count);                \
4801         if (__data < (MIN))                                             \
4802                 __data = (MIN);                                         \
4803         else if (__data > (MAX))                                        \
4804                 __data = (MAX);                                         \
4805         *(__PTR) = (u64)__data * NSEC_PER_USEC;                         \
4806         return ret;                                                     \
4807 }
4808 USEC_STORE_FUNCTION(cfq_slice_idle_us_store, &cfqd->cfq_slice_idle, 0, UINT_MAX);
4809 USEC_STORE_FUNCTION(cfq_group_idle_us_store, &cfqd->cfq_group_idle, 0, UINT_MAX);
4810 USEC_STORE_FUNCTION(cfq_slice_sync_us_store, &cfqd->cfq_slice[1], 1, UINT_MAX);
4811 USEC_STORE_FUNCTION(cfq_slice_async_us_store, &cfqd->cfq_slice[0], 1, UINT_MAX);
4812 USEC_STORE_FUNCTION(cfq_target_latency_us_store, &cfqd->cfq_target_latency, 1, UINT_MAX);
4813 #undef USEC_STORE_FUNCTION
4814
4815 #define CFQ_ATTR(name) \
4816         __ATTR(name, S_IRUGO|S_IWUSR, cfq_##name##_show, cfq_##name##_store)
4817
4818 static struct elv_fs_entry cfq_attrs[] = {
4819         CFQ_ATTR(quantum),
4820         CFQ_ATTR(fifo_expire_sync),
4821         CFQ_ATTR(fifo_expire_async),
4822         CFQ_ATTR(back_seek_max),
4823         CFQ_ATTR(back_seek_penalty),
4824         CFQ_ATTR(slice_sync),
4825         CFQ_ATTR(slice_sync_us),
4826         CFQ_ATTR(slice_async),
4827         CFQ_ATTR(slice_async_us),
4828         CFQ_ATTR(slice_async_rq),
4829         CFQ_ATTR(slice_idle),
4830         CFQ_ATTR(slice_idle_us),
4831         CFQ_ATTR(group_idle),
4832         CFQ_ATTR(group_idle_us),
4833         CFQ_ATTR(low_latency),
4834         CFQ_ATTR(target_latency),
4835         CFQ_ATTR(target_latency_us),
4836         __ATTR_NULL
4837 };
4838
4839 static struct elevator_type iosched_cfq = {
4840         .ops = {
4841                 .elevator_merge_fn =            cfq_merge,
4842                 .elevator_merged_fn =           cfq_merged_request,
4843                 .elevator_merge_req_fn =        cfq_merged_requests,
4844                 .elevator_allow_bio_merge_fn =  cfq_allow_bio_merge,
4845                 .elevator_allow_rq_merge_fn =   cfq_allow_rq_merge,
4846                 .elevator_bio_merged_fn =       cfq_bio_merged,
4847                 .elevator_dispatch_fn =         cfq_dispatch_requests,
4848                 .elevator_add_req_fn =          cfq_insert_request,
4849                 .elevator_activate_req_fn =     cfq_activate_request,
4850                 .elevator_deactivate_req_fn =   cfq_deactivate_request,
4851                 .elevator_completed_req_fn =    cfq_completed_request,
4852                 .elevator_former_req_fn =       elv_rb_former_request,
4853                 .elevator_latter_req_fn =       elv_rb_latter_request,
4854                 .elevator_init_icq_fn =         cfq_init_icq,
4855                 .elevator_exit_icq_fn =         cfq_exit_icq,
4856                 .elevator_set_req_fn =          cfq_set_request,
4857                 .elevator_put_req_fn =          cfq_put_request,
4858                 .elevator_may_queue_fn =        cfq_may_queue,
4859                 .elevator_init_fn =             cfq_init_queue,
4860                 .elevator_exit_fn =             cfq_exit_queue,
4861                 .elevator_registered_fn =       cfq_registered_queue,
4862         },
4863         .icq_size       =       sizeof(struct cfq_io_cq),
4864         .icq_align      =       __alignof__(struct cfq_io_cq),
4865         .elevator_attrs =       cfq_attrs,
4866         .elevator_name  =       "cfq",
4867         .elevator_owner =       THIS_MODULE,
4868 };
4869
4870 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4871 static struct blkcg_policy blkcg_policy_cfq = {
4872         .dfl_cftypes            = cfq_blkcg_files,
4873         .legacy_cftypes         = cfq_blkcg_legacy_files,
4874
4875         .cpd_alloc_fn           = cfq_cpd_alloc,
4876         .cpd_init_fn            = cfq_cpd_init,
4877         .cpd_free_fn            = cfq_cpd_free,
4878         .cpd_bind_fn            = cfq_cpd_bind,
4879
4880         .pd_alloc_fn            = cfq_pd_alloc,
4881         .pd_init_fn             = cfq_pd_init,
4882         .pd_offline_fn          = cfq_pd_offline,
4883         .pd_free_fn             = cfq_pd_free,
4884         .pd_reset_stats_fn      = cfq_pd_reset_stats,
4885 };
4886 #endif
4887
4888 static int __init cfq_init(void)
4889 {
4890         int ret;
4891
4892 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4893         ret = blkcg_policy_register(&blkcg_policy_cfq);
4894         if (ret)
4895                 return ret;
4896 #else
4897         cfq_group_idle = 0;
4898 #endif
4899
4900         ret = -ENOMEM;
4901         cfq_pool = KMEM_CACHE(cfq_queue, 0);
4902         if (!cfq_pool)
4903                 goto err_pol_unreg;
4904
4905         ret = elv_register(&iosched_cfq);
4906         if (ret)
4907                 goto err_free_pool;
4908
4909         return 0;
4910
4911 err_free_pool:
4912         kmem_cache_destroy(cfq_pool);
4913 err_pol_unreg:
4914 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4915         blkcg_policy_unregister(&blkcg_policy_cfq);
4916 #endif
4917         return ret;
4918 }
4919
4920 static void __exit cfq_exit(void)
4921 {
4922 #ifdef CONFIG_CFQ_GROUP_IOSCHED
4923         blkcg_policy_unregister(&blkcg_policy_cfq);
4924 #endif
4925         elv_unregister(&iosched_cfq);
4926         kmem_cache_destroy(cfq_pool);
4927 }
4928
4929 module_init(cfq_init);
4930 module_exit(cfq_exit);
4931
4932 MODULE_AUTHOR("Jens Axboe");
4933 MODULE_LICENSE("GPL");
4934 MODULE_DESCRIPTION("Completely Fair Queueing IO scheduler");